Google ドキュメントを使用して公開済み
ICU Meeting Minutes 2008
5 分ごとに自動更新

ICU Meeting Minutes 2008 (Archived)

Current Meeting Minutes

2008/12/24 and 2008/12/31- no meeting

2008/12/17

  1. Note:  callin # will change - Steven to post updated number.  Check back on the agenda before future meetings. (srl)
  2. User Guide on Google Site (Markus)
  1. Include lenient (collation) rules in RBNF? (mark, john)
  1. English RBNF rules  search for %%lenient
  1. & ':' = '.' = ' ' = '-';
  1. http://source.icu-project.org/repos/icu/icu/trunk/source/data/rbnf/
  2. very sparsely populated right now.
  3. John to leave as ICU special
  4. Already done in trunk.
  1. To review this time: ICU @provider interface - Ticket #6631 (Steven)
  1. http://docs.google.com/Doc?id=ddjsbnb9_13cqqtk6df
  2. Q: This adds a global keyword that affects other services.  This would be the first time that more than one keyword applies to a service.
  3. If this is plugged in, then "provider" needs to be added to the list of keyword
  4. by default, behavior would not be there in 4.2 (in future, maybe dynamic loader)
  5. Java: No plan for this in icu4j for now. However, with the JDK, may use  "@provider" keyword within the JDK:   "de@provider=JDK" vs "de@provider=ICU"
  1. API(s) for getting info from supplemental: language-related info (e.g. primary territories), territory-related info (e.g. calendars, telephone codes) - Peter's design proposal of 12/15 which Mark wanted to review in the meeting, subject "[icu-design] API to get likely territories and/or scripts for language code"
  1. put in ulocdata.h
  2. not a static method
  3. API options:
  1. always use locale, pack with und_ZZ -  so if you want information about AX,  pass in locale code "und_AX" to request
  2. take in 'territory codes' as API
  1. pass in full locale, just use relevant pieces?
  2. use case: setup assistant
  1. given language, list of countries based on language (with 'show all…')
  1. May want to broaden from 'primary/secondary'. See supplementalInfo.
  2. Peter to revisit.
  1. Main site: http://sites.google.com/site/icusite/ (mark)
  1. Email discussion of pages to port: subject "[icu-core] scraper implemented"
  2. Mark to port pages over
  3. Mark and Markus to collect redirects in separate document, and link from agenda.
  4. redirects - collect here: http://docs.google.com/Doc?id=ddsrrpj5_51fh7j2xfg
  1. String Search
  1. how to handle han[gul]
  2. Han CEs handled as implicit
  1. Bugs -next time.
  2. Kedar
  3. Next meeting- jan 7th  (ICU 4.0.1 GA jan 9th- communication via email.)

2008/12/10

  1. Time zone formatting/parsing improvement plan (Yoshito)
  1. David, Andy, Peter, Eric and Yoshito had a meeting on Dec 9.
  2. Note and meeting log -> https://docs.google.com/Doc?id=d3q3z35_17c9t2gxcn 
  3. Agreed to refactor some ICU time zone resource structure (Peter)
  4. Lazy instantiation of full time zone display name array/trie (Yoshito)
  1. Proposed 4.0.1/4.2M2 plan update (Steven)
  1. Request from an IBM internal consumer - B-M string search implementation in ICU4C
  1. prototype in as separate API - marked @internal, tech preview
  2. as well as some minor fixes- reverse CEIter, offsets were broken before
  3. J not affected
  1. Synchronized J release with C release
  2. Postpone 4.0.1 GA to Jan 9, 2009 (Release Candidate Jan 6th, 2009)
  3. Push 4.2M2 out 2 weeks (Jan 7 to Jan 21)
  4. Steven to update schedule/file bugs
  5. J fix list sent out (Yoshito)
    C list needs to be sent out (Steven todo)
  1. Google Sites for manually edited ICU web pages? (not User Guide, API references, Trac) (Mark)
  1. Mark experimentally created http://sites.google.com/site/icusite/
  2. Mark asked Steven to set up site.icu-project.org with a CNAME as per the instructions to try it out
  3. Moving forward with more thorough test.
  4. Try to minimize difference in appearance. But not a huge issue.
  1. Simplify User Guide editing?! (Andy)
  1. For example, turn the User Guide chapters into pages and subpages on the Google Site for WYSIWYG editing
  2. Question: Do we need to generate the PDF file? If not, then we don't need an elaborate (OOo or XML) source form
  3. Question: Do we need to be able to stage edits, or is it ok to make them go live immediately?
  4. For release downloads of the User Guide, we can probably write a small scraper tool that downloads all of its pages and subpages
  1. Online code reviews (Markus)
  1. Markus added notes in http://bugs.icu-project.org/trac/ticket/6649
  2. Michael installed the Trac PeerReview plug-in -- judging from the screenshots, it's a little clunkier to use than Rietveld, but much easier to install since it uses Trac's infrastructure (e.g., its database)
  3. Markus (and others?) to try out PeerReview
  4. Markus to do a little more reading about available code review tools
  1. Exceptions in Java (mark)
  1. Add package-private subclasses of the common Exceptions we throw inside ICU.
  2. Add ICUException as interface that these all use.
  3. Allows for debugging use, and for programs to catch ICU exceptions separately.
  4. Mark to file bug.
  1. Bugs !
  2. To review next time: ICU @provider interface - Ticket #6631 (Steven)
  1. http://docs.google.com/Doc?id=ddjsbnb9_13cqqtk6df
  1. No meeting Dec 24th

2008/12/3

  1. Bugs !
  2. 4.0 maintenance release - 4.0.1 (Yoshito)
  1. TimeZone.getDisplayName problem was actually reproduced by ICU4J too (#6644)
  2. Yoshito proposed to have 4.0.1 for C and J.
  3. Let's review the candidate list...
  4. Dec 10th: 4.0.1 final bug list
  5. Dec 17th:  4.0.1 release date
  1. Who uses ICU? (http://icu-project.org/ ) (Mark)
  1. We should have an anchor right to that section
  2. We should restructure the list to first have all the companies (including IBM). That way a copy of the companies doesn't miss IBM. Then we could have the list of IBM products, followed by (if they want) product lists from other core members.
  3. Add NetApp
  4. Mark to make changes. 
  1. Hosting the site on Sites?
  1. Maybe using Google, let's see if we can get it on our domain.
  1. Yes, we would set site.icu-project.org or something, and the root page could redirect there.
  2. Review next week.
  1. TimeZone memory requirements, time zone formatting/parsing performance issues (Peter, Yoshito)
  1. >100K for CanonicalMap, OlsonToMeta, MetaToOlson; ticket #6645
  1. First try to optimize internal data structures, e.g., try using UChar* instead of UnicodeString objects
  1. ticket #6426 - large trie built for timezone parsing
  2. Currently putting all timezone names for locale into one big trie, including all variants
  3. A year ago, Mark wrote down parsing logic that did not require such a large trie
  4. Already smaller, 6414 (was 1.4M, now 180K)?
  1. should be done on parse, not when formatter created
  2. it is per language+territory+etc
  3. many users may only use a small number of zones
  1. have language-neutral data split out
  2. Will come up with a proposal and separate call
  3. Separate discussion group on this: Yoshito, Peter, Andy, Dave, Mark
  1. Yoshito will schedule a meeting: Dec 9, 2 PM PT
  2. Everyone should investigate approaches to improve this

2008/11/26

2008/11/19

  1. Available keyword values for a locale in a service(getKeywordValuesForLocale) (Kedar)
  1. Mark> For Java, we should probably have an Interface to encompass these APIs (we'll have to look back at your Java proposal).
  2. Steven> Hmm, commonlyUsed may be 'preferred' vs. 'available' as per CLDR.
  3. User wants:
  1. preferred list - being added to cldr.  An ordered list of what is preferred.
  2. commonly used / customary
  3. "what makes a difference"  - 
    en@collation=phonebook == en@collation=dictionary,
    de@collation=phonebook ≠ de@collation=dictionary
  4. "what is actually available/supported" - chinese calendar in russian may have english data
  1. What kinds
  1. calendar
  2. currency
  3. collator
  4. timezone?
  1. We can merge 'preferred' with 'customary' - only list customary calendars in the preferred list. ( a 'secondary' list could list non-customary calendar order? )
  2. 'what is supported' is hard to answer, may or may not need display. Don't do.
  3. 'what makes a difference' can be done by looking at the fallback chain.
  4. Potentially two lists:  'customary, in preferred order', and 'what makes a difference'
  1. customary, in preferred order
  1. currency: only currently active currency
  1. what makes a difference
  1. [all currency]
  2. [all calendars]
  1. verify that each service has an API for, and (CLDR) data is available for, and the documentation is correct for:
  1. all available keywords
  2. and all available values.
  3. customary in preferred order, verify that 'default' is the first item. 
  1. calendar: ok
  2. collator: no data in cldr
  3. currency : verify that we have this list
  4. timezone :
  1. Issue: what if this is a language only-locale?  i.e. preferred currency for 'en'. We will apply addLikelySubtags.
  1. Document that this is 'best available' data.  Default in the absence of other data. Different applications may have different preference orders.
  2. Also, '_DE' may expand to 'de_DE', including phonebook.
  1. OpenJDK Locale Enhancement project (yoshito,doug)
  1. http://www.openjdk.org/projects/locale-enhancement/ 
  2. Mailing list - locale-enhancement-dev -> http://mail.openjdk.java.net/mailman/listinfo/locale-enhancement-dev
  1. ICU resource bundle issues (yoshito)- re: "[icu-core] ICU resource bundle issues 11/5/08"
  1. Some design oriented issues
  1. Multi-level fallback (getByKeyWithFallback)
  2. ICU4J UResourceBundle as a Java ResourceBundle subclass?
  3. Available keys in a table
  1. Yoshito's "rough" proposal for ICU4J
  1. Separating low level data access APIs from user APIs.
  2. Deprecate UResourceBundle
  3. Create a new ResourceBundle class, but not as a subclass of java.util.ResourceBundle
  4. Create a new java.util.ResourceBundle subclass - ResourceCollection. which only support key-value mapping for a single bundle.  No locale inheritance is considered in this class.
  5. new ICU ResourceBundle#getBundle will initialize multiple ResourceCollection (each one is corresponding to a .res file) and build a locale inheritance chain.
  6. The new ICU ResourceBundle give you a view of logical resource tree (resources from multiple .res are merged)
  7. In the new ICU ResourceBundle, all get methods act like [currently internal] getByKeyWithFallback (fallback is always involved / can specify hierarchical key)
  8. ResourceCollection will parse the contents of .res and build the run-time friendly structure.  The raw byte[] contents will be released.
  1. Discuss Multi Level Fallback:
  1.  Original design: Java/Windows.  get 'Dialog Box' as a single structured item, with nested sub items. Would not want it to partially fall back.
  2. 'array' does not fall back- but a dialog may be key/value pairs
  3. don't want to break this behavior. can add API that will do this fallback, and hierarchical key [ get("calendars/gregorian/...")  ]
  4. could have a syntax to handle this:   dateFormatSymbols/4/…
  5. CLDR has some items which are ordered, such as collator contents.
  1. Discuss ICU4J UResourceBundle vs Java ResourceBundle  and proposal
  1. ResourceCollection (separate obj for [root+sr+sr_Latn]
  1. if there are aliases that point outside of your 'parent chain', does complicate things.
  2. should have all context in the bundle.
  1. currently evaluating prototype. 
  2. 'unflatten all at load time'  - #8 somewhat independent.
  1. historically, didn't do it due to concern about (1) load time, and (2) small % of items that are typically used. However, no measurement was done.
  1. Yoshito to write up more concrete proposal, motivation, and example
  1. make google doc, link here, link to bug(s)
  2. srl to look at C API issues.
  1. Q:  We have 'createXXXX(ResourceBundle)' and 'createXXXX(UResourceBundle)' - will we need new overloads?
  1. very few public APIs..
  1. Q: desync between C and J ?
  1. talk about corresponding changes
  2. the J ResourceBundle is a new API
  1. Q: we still want it to be a ResourceBundle?  Yoshito to answer.. design issue.
  1. most of these changes should be internal.
  1. Q: could document that keys containing slash in getByKey() do multilevel lookup and fallback
  1. already have lots of get*byKey. 
  2. slash is already a prohibited key name
  1. ex> supplemental data has:            "meta:Australia_Central_001"{"Australia/Adelaide"}
  1. drawback: unexpected behavior with addition of a slash,  "key/"
  2. look at the API.  top level vs. sublevel fallback.
  1. Bugs !

2008/11/12 —   Back to 11am PT meeting time! 

  1. Locale Canonicalization (Steven?)
  1.  6593 - bug
  2. hard coded mappings are in C and J:  IANA/BCP47, .NET, ICU compatibility, …
  3. icu/source/data/*.xml contain compat mapping that generate alias locales (de__PHONEBOOK, etc)
  4. Steven to file CLDR bug
  1. CLDR / tool status? (John)
  1. RBNF (John)
  1. Proposal to change the structure in the resource bundles keeping the RBNF syntax the same
  2. en { // current
         SpelloutRules {
            "%simplified:\n"
                "-x: minus >>;\n"
            "%default:\n"
                "-x: minus >>;\n"
                "x.x: << point >>;\n"
        }
        OrdinalRules {
            "%main:\n"
                "=#,##0==%%abbrev=;\n"
            "%%abbrev:\n"
        }
    }
  3. en { // proposed
         RBNFRules {
            "%spellout:\n",
                "-x: minus >>;\n",
            "%default:\n",
                "-x: minus >>;\n",
                "x.x: << point >>;\n",
            "%ordinal:\n"
                "=#,##0==%%abbrev=;\n"
            "%%abbrev:\n"
    }
  4. en {
             RBNFRules {
                  simplified {
                      "-x: minus >>;\n"
                  }
                  default {
                      "-x: minus >>;\n".
                      "x.x: << point >>;\n"
                  }
                  ordinal {
                      "=#,##0==%%abbrev=;\n"
                  "%%abbrev:\n"
                  }
            }
    }
  5. Rule Structure flattening (srl)
  1. old:
  1. Ordinal
  1. default
  2. x
  3. y
  4. z
  1. Spellout
  1. default
  2. x
  3. y
  4. z
  1. new
  1. ordinal
  2. ordinal_x
  3. ordinal_y
  4. ordinal_z
  5. spellout
  6. spellout_x
  7. spellout_y
  8. spellout_z
  9. checkwriting
  10. …
  1. Proposal is http://www.unicode.org/cldr/data/dropbox/misc/RBNFRules.html
  2. Concern with flattening (Doug)
  1. predefined sets
  2. behavior concerning using the last public ruleset
  3. could remove restriction on fixed set of keys
  4. multiple (private?) variations per key: ex, hebrew. 
  5. reimplement before defining CLDR data
  1. Difficulty: how to name sets across locales.
  2. Fallback from ordinal_foobar to ordinal
  3. John TODO: Comfortable with #4 above. Let's see how current rules would map to this format. 
  4. John to set up separate meeting to review RBNF design.
  1. ICU resource bundle issues (yoshito)
  1. Multi-level fallback (getByKeyWithFallback)
  2. ICU4J UResourceBundle as a Java ResourceBundle subclass?
  3. Available keys in a table
  4. Numeric index access to table items
  5. Sent a message to icu-core Nov 5th, please read for discussion 11/19.
  1. Available keyword values for a locale in a service(getKeywordValuesForLocale)
  1. Mark> For Java, we should probably have an Interface to encompass these APIs (we'll have to look back at your Java proposal).
  2. Steven> Hmm, commonlyUsed may be 'preferred' vs. 'available' as per CLDR.
  1. Bugs !

2008/11/05

  1. Meeting time? (Mark)
  1. Agreement: move meeting back to 11am PT
  1. LayoutEngine isBogus() (Eric/Jason)
  1. isBogus is new (not in released ICU)
  2. preferred model is to use the layout engine factory- which has always (?) had an error code.
  3. Public c'tor is already @internal: (LayoutEngine in 4.0) - users shouldn't be calling it directly
  4. Add error code to c'tor and all subclasses, change factory to use the version with the error code, remove isBogus
  1. Yoshito: add #6612 (ICU4J charset buffer bug) to 4.0.1 maintenance candidate list
  2. incoming bugs
  1. 6613 - srl to fix, peter to file new rfe
  2. 6599 - doug holiday
  3. 6603 - part of bcp functionality. Dave to document issues. CC: yoshito. M4.2
  4. 6609 - toPattern()/toLocalizedPattern(). yoshito to reply with comments and wontfix.
  5. 6611 - doug to look at content,  srl to check with IBM legal, possibly put an implicit contribution note on the bug submission form
  6. 6614 - peter to do.  4.1.2
  7. 6615 - peter to do.  cc: srl. 4.1.2 TRADITIONAL / GREGORIAN . UCalendarType
  1. UCAL_TRADITIONAL = UCAL_DEFAULT, no change (current behavior)
  2. UCAL_GREGORIAN: internally  calls setKeywordValue("calendar", "gregorian") - > @calendar=gregorian
  3. "ja_JP", UCAL_TRADITIONAL -> gregorian.  "no change"
  1. ICU Resource Bundle issues (Yoshito) - discuss next week.
  2. configure / ^M woes (Steven/Markus)
  1. if you touch configure/.in let srl know..
  2. autoconf is going to remove the bare ^M in a future version.
  1. getDisplayName bug in 4.0 - 6597 - errata page on 4.0?  (Steven/Yoshito)
  1. The fix was merged back to maint-4-0 - http://bugs.icu-project.org/trac/changeset/24940 
  2. ICU 4.0 download page was updated - http://source.icu-project.org/download/4.0.html#errata-2 
  3. Yoshito sent a message about the issue to icu-support/announce/design list on Nov 7, 2008

2008/10/29

  1. Broken Currency choice formats in ICU 4.0.  The problem is bad CLDR data.  Affects az, de, fil, fr, ko, th, vi, zh_Hant.  Mostly INR.  Also something funny with hu, bad in ICU, ok in CLDR.  (Andy, Doug)
  1. TZ bugs (Yoshito)
  1. 6597 - C GetDisplayName - done in trunk
  2. 6602 (cldrbug:1962)  new zone in 2008i - spec change in CLDR, need code change in ICU. Yoshito to file and number. John to file CLDR bug.
  1. New Google ICU committers: Fredrik Roubert [roubert] (Trondheim) & Shaopeng Jia [shaopengjia] (Zürich)
  2. Incoming Bugs
  1. Steven to file bug about testing with fake "current time", continuous monster test for dates. Done.

2008/10/22

  1. Reviews- thanks, more to do (see below) (srl)
  2. Proposal for encoding conversion security issues TR (Peter)
  1. would be good to have ICU samples in a separate ICU doc that refers to the Unicode TR
  2. Unicode TR will just have pseudocode
  3. Peter will make a proposal at next UTC meeting
  1. http://source.icu-project.org/repos/icu/data/trunk/meta/ - ICU Metadata project (Steven)
  1. Information about versions of ICU and their release dates
  2. Who has tested which version of ICU on which platforms
  3. Potentially auto-generate pieces of the download page
  4. ->  http://icu-project.org/xml/  and http://icu-project.org/dtd/
  1. Incoming Bugs

2008/10/15

  1. Converters toUnicode handling of illegal sequences, #5691 (Markus, Jungshik, Erik)
  1. Code point ranges: start+limit or start+end (see email) (Markus)
  1. Bug Cleanup, 4.1.1 (srl)
  1. 4.1.1, but no owner -> #6447 TODO: Google to figure out priority & scheduling & owner
  2. 4.1.1, very old code, but no reviewer -> #5503 , #5869
  3. 4.1.1, open and no reviewer: [query] - Please either assign reviewer or move to 4.1.2
  4. 4.1.1, has a reviewer: [query] - please complete reviews!
  5. Unscheduled, Active tickets with code and no reviewer: [query]
  6. 4.0 bug reviews-  please complete review or assign another reviewer! [query]
  1. Incoming Bugs

2008/10/8

  1. [moved to next meeting] Bug Cleanup, 4.1.1
  1. 4.1.1, but no owner -> #6447
  2. 4.1.1, very old code, but no reviewer -> #5503 , #5869
  3. 4.1.1, open and no reviewer: [query] - Please either assign reviewer or move to 4.1.2
  4. 4.1.1, has a reviewer: [query] - please complete reviews!
  5. Unscheduled, Active tickets with code and no reviewer: [query]
  6. 4.0 bug reviews-  please complete review or assign another reviewer! [query]
  1. tzdata 2008g
  1. ticket #6577
  2. for now- works through 2038, but need to add 64 bit (pre 1901, post 2038) - #5624
  3. done for C, not J yet-
  1. 64bit libraries in C (srl) - #6564 
  1. proposal: deprecate default behavior, replace with --with-library-bits=32/64
  1. Warnings on by default in C (srl)
  1. shadowing of index()  -Wshadow -?
  1. index is not one of the parameter names recommended by coding guidelines
  1. Umbrella tickets listed under milestone:4.2  - such as code cleanup for C
  2. ERA: #6547 - #define ERA,  langinfo.h selector for nl_langinfo()
  1. Incoming Bugs
  2. ustdio.h still has functions with "@draft 3.0" - should they become stable? (Peter)
  1. yes, @stable 3.0 - Peter to file bug → #6578

2008/10/1

  1. Data Thoughts:  would it be reasonable to make available with ICU releases a copy of the CLDR data that exactly corresponds to the included ICU data?  We at Google would find this convenient.  (andy)
  1. Time Zone Data Updates.  Do I assume correctly, in the case of metazoneinfo changes, that the ICU files will be the first to be fixed, and the CLDR xml files will follow?  (andy)
  1. autoconf problems. configure.in prereqs version 2.63, which was released by the GNU people 9 Sept 2008.  Suffice it to say that NO system distributions yet include this version, it must be built from sources. Do we really need this version? (I was trying to switch the compiler options to enable strict warnings.) (andy)
  1. code reviews (srl): Google tardy reviewing...
  2. 4.1.1 release - milestone-4-1-1 (srl, yoshito)
  1. bugs Incoming Bugs
  2. Doug sent out JDK Locale Enhancement note to openjdk announce list, to propose a new Project.
  3. Q: should milestones change

                

2008/9/24

  1. Discuss M1 pending changes, announce proposed svn revision number for C and J M1
  1. John to finish collation data relocation in ICU4J
  2. Fire full build/testing for C now.  Wait for the collation change above for J.
  1. Request for participating ICU PMC from Benedicto Franco Jr. (yahoo)
  1. PMC membership is not based on company.  Tex was representing yahoo before, but it does not mean we'll fill in the position automatically.
  2. We agreed to follow following steps
  1. See if Benedicto can actually contribute things to the project, such as
  1. Active involvement in ICU mailing lists - design suggestions, proposals, comments...
  2. Contribution of new features/improvements/bug fixes
  3. Statement from his management addressing his commitment to the ICU project (for example, work time which he can use for the project) may help.
  1. After we confirm his contributions, we'll give him a committer access to the ICU repository.
  2. Then, we will vote for his involvement in the PMC
  1. Milestone BRS task assignment
  1. Steven/Yoshito will work on most of items.
  2. May ask someone's help on some tasks.
  3. If someone is willing to volunteer on some tasks, send a message to Steven/Yoshito
  1. Quick question about adding RuleBasedBreakIterator constructor that does not adopt data (Peter): icu-design mail 9/18, ticket #5858
  1. Agreed to add the option.  enum maybe used instead of boolean
  1. Resource bundle's behavior when fallback and alias to /LOCALE is involved.
  1. Yoshito posted the design question in the core ML on 9/24
  2. Yoshito to come up with more concrete examples and discuss this again in the next PMC.
  1. bugs Incoming Bugs

2008/9/17 (change to new meeting time 10:00 AM PT)

  1. New ICU tools work (Brian)
  1. 6353 - LDML2ICUConverter output .res directly ( minus Collator binary, some transliterator files )
  2. 4966 - ICU2LDML
  1. 4.0 cleanup
  1. http://bugs.icu-project.org/trac/ticket/5696 (Assigned to: markus)
    http://bugs.icu-project.org/trac/ticket/5415 (Assigned to: mark)
  2. 25 total remaining reviews for 4.0 stream: http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&groupdesc=1&group=revw&milestone=4.0&milestone=3.9.1&milestone=3.9.2&milestone=3.9.3&order=priority
  1. bugs Incoming Bugs

2008/9/12 (10:30 AM PT) - meeting moved from 9/10 due to IUC

  1. bugs Incoming Bugs
  2. Java Locale Update (Yoshito)
  1. proposed enhancements to JDK- so Java supports BCP47
  2. suggestion: start OpenJDK Project (under Internationalization Group)
  3.   Doc URL: http://docs.google.com/Doc?id=d3q3z35_10g9wwphg6
  4. want to put some implementation into ICU, as ICU will need to depend on similar functionality
  1. M1 small BRS (Oct 1)
  1. agreement was: make milestones more stable/usable.
  2. http://icu-project.org/processes/release.html 
  3. some items are a mess at the end, so do them at each point
  1. scan for copyright
  2. no-data test -
  1. (check srl's crash with custom data from iuc32)
  2. do, and tell people who worked on the code to be more careful
  1. bug reviews
  2. environment test
  3. uconfig test
  4. compile warnings
  1. andy to look at increasing the base warning level


ICU Meeting Minutes
If you are not editing this document, please just view it with Preview (under Share). You'll need to refresh when a change is made.For historical minutes prior to 2007/04/25, see http://www.icu-project.org/meetings/.Action Items

  1. Number

  1. Due
  1. Done
  1. Who
  1. Description

Completed items are moved to the table at the bottom of this document, except for the last one, which is left here so that we have the last number.
ICU 4.0.1 worksheet moved to new doc >> Future Topics

  1. ICU Schedule
  1. ICU 4.0.1 Schedule
  1. ICU 4.2 Schedule

  1. 2009/01/14
  1. ICU 4.0.1 status (Steven, Yoshito)
  2. ICU 4.1.2 schedule - M2 upcoming next Wed (1/21)
  3. ICU 4.2 Platforms
  1. C/J: IBM binary/testing plans (Steven, Yoshito)
  2. C: VC 2005 -> 2008?
  3. IBM test:
  1. ICU4C
    Windows Vista (32-bit), Visual Studio 2005 2008 express

    Windows Server 2003 2008 (64-bit), Visual Studio 2005 2008AIX 6.1 VisualAge 9 (64-bit)
    Solaris 10 (64-bit), Sun Studio 12
    Red Hat Enterprise Linux 5 (32-bit), GCC 4.1.2
    '''+Red Hat Enterprise Linux 5 (64-bit), GCC 4.1.2
    HP 11iv3 IA64 (64-bit) - aCC A.6.14
  2. ICU4J

  1. Note: Class files are Java 1.3 compatible (javac -target 1.3)
  1. Reminder - closing milestone 4.0 / 4.1.1
  1. We still need several tickets reviewed for 4.0
  1. Bugs

  1. 2009/01/07
  1. ICU 4.0.1 status (Steven/Yoshito)
  1. Steven and Yoshito to update
  1. C++ static cast (Andy)
  1. Andy to try some in a branch - IBM team tries it with various compilers
  2. If it works, we will use them in new code.. Probably not trying to update everything at one time. 
  1. ICU Numbering System proposal (John)
  1. John to send out the proposal shortly.
  2. Force latin supported?
  1. The proposal is more comprehensive
  2. Yes? - Can be implemented like this way.
  3. How can we handle alternate decimal symbols (decimal point, grouping separator), not only for digits (0-9)?
  1. Bugs

  1. 2008/12/24 and 2008/12/31- no meeting
    2008/12/17
  1. Note:  callin # will change - Steven to post updated number.  Check back on the agenda before future meetings. (srl)
  2. User Guide on Google Site (Markus)
  1. Include lenient (collation) rules in RBNF? (mark, john)
  1. English RBNF rules  search for %%lenient
  1. & ':' = '.' = ' ' = '-';
  1. http://source.icu-project.org/repos/icu/icu/trunk/source/data/rbnf/
  2. very sparsely populated right now.
  3. John to leave as ICU special
  4. Already done in trunk.
  1. To review this time: ICU @provider interface - Ticket #6631 (Steven)
  1. http://docs.google.com/Doc?id=ddjsbnb9_13cqqtk6df
  2. Q: This adds a global keyword that affects other services.  This would be the first time that more than one keyword applies to a service.
  3. If this is plugged in, then "provider" needs to be added to the list of keyword
  4. by default, behavior would not be there in 4.2 (in future, maybe dynamic loader)
  5. Java: No plan for this in icu4j for now. However, with the JDK, may use  "@provider" keyword within the JDK:   "de@provider=JDK" vs "de@provider=ICU"
  1. API(s) for getting info from supplemental: language-related info (e.g. primary territories), territory-related info (e.g. calendars, telephone codes) - Peter's design proposal of 12/15 which Mark wanted to review in the meeting, subject "[icu-design] API to get likely territories and/or scripts for language code"
  1. put in ulocdata.h
  2. not a static method
  3. API options:
  1. always use locale, pack with und_ZZ -  so if you want information about AX,  pass in locale code "und_AX" to request
  2. take in 'territory codes' as API
  1. pass in full locale, just use relevant pieces?
  2. use case: setup assistant
  1. given language, list of countries based on language (with 'show all…')
  1. May want to broaden from 'primary/secondary'. See supplementalInfo.
  2. Peter to revisit.
  1. Main site: http://sites.google.com/site/icusite/ (mark)
  1. Email discussion of pages to port: subject "[icu-core] scraper implemented"
  2. Mark to port pages over
  3. Mark and Markus to collect redirects in separate document, and link from agenda.
  4. redirects - collect here: http://docs.google.com/Doc?id=ddsrrpj5_51fh7j2xfg
  1. String Search
  1. how to handle han[gul]
  2. Han CEs handled as implicit
  1. Bugs -next time.
  2. Kedar
  3. Next meeting- jan 7th  (ICU 4.0.1 GA jan 9th- communication via email.)


  1. 2008/12/10

  1. Time zone formatting/parsing improvement plan (Yoshito)
  1. David, Andy, Peter, Eric and Yoshito had a meeting on Dec 9.
  2. Note and meeting log -> https://docs.google.com/Doc?id=d3q3z35_17c9t2gxcn 
  3. Agreed to refactor some ICU time zone resource structure (Peter)
  4. Lazy instantiation of full time zone display name array/trie (Yoshito)
  1. Proposed 4.0.1/4.2M2 plan update (Steven)
  1. Request from an IBM internal consumer - B-M string search implementation in ICU4C
  1. prototype in as separate API - marked @internal, tech preview
  2. as well as some minor fixes- reverse CEIter, offsets were broken before
  3. J not affected
  1. Synchronized J release with C release
  2. Postpone 4.0.1 GA to Jan 9, 2009 (Release Candidate Jan 6th, 2009)
  3. Push 4.2M2 out 2 weeks (Jan 7 to Jan 21)
  4. Steven to update schedule/file bugs
  5. J fix list sent out (Yoshito)
    C list needs to be sent out (Steven todo)
  1. Google Sites for manually edited ICU web pages? (not User Guide, API references, Trac) (Mark)
  1. Mark experimentally created http://sites.google.com/site/icusite/
  2. Mark asked Steven to set up site.icu-project.org with a CNAME as per the instructions to try it out
  3. Moving forward with more thorough test.
  4. Try to minimize difference in appearance. But not a huge issue.
  1. Simplify User Guide editing?! (Andy)
  1. For example, turn the User Guide chapters into pages and subpages on the Google Site for WYSIWYG editing
  2. Question: Do we need to generate the PDF file? If not, then we don't need an elaborate (OOo or XML) source form
  3. Question: Do we need to be able to stage edits, or is it ok to make them go live immediately?
  4. For release downloads of the User Guide, we can probably write a small scraper tool that downloads all of its pages and subpages
  1. Online code reviews (Markus)
  1. Markus added notes in http://bugs.icu-project.org/trac/ticket/6649
  2. Michael installed the Trac PeerReview plug-in -- judging from the screenshots, it's a little clunkier to use than Rietveld, but much easier to install since it uses Trac's infrastructure (e.g., its database)
  3. Markus (and others?) to try out PeerReview
  4. Markus to do a little more reading about available code review tools
  1. Exceptions in Java (mark)
  1. Add package-private subclasses of the common Exceptions we throw inside ICU.
  2. Add ICUException as interface that these all use.
  3. Allows for debugging use, and for programs to catch ICU exceptions separately.
  4. Mark to file bug.
  1. Bugs !
  2. To review next time: ICU @provider interface - Ticket #6631 (Steven)
  1. http://docs.google.com/Doc?id=ddjsbnb9_13cqqtk6df
  1. No meeting Dec 24th


  1. 2008/12/3

  1. Bugs !
  2. 4.0 maintenance release - 4.0.1 (Yoshito)
  1. TimeZone.getDisplayName problem was actually reproduced by ICU4J too (#6644)
  2. Yoshito proposed to have 4.0.1 for C and J.
  3. Let's review the candidate list...
  4. Dec 10th: 4.0.1 final bug list
  5. Dec 17th:  4.0.1 release date
  1. Who uses ICU? (http://icu-project.org/ ) (Mark)
  1. We should have an anchor right to that section
  2. We should restructure the list to first have all the companies (including IBM). That way a copy of the companies doesn't miss IBM. Then we could have the list of IBM products, followed by (if they want) product lists from other core members.
  3. Add NetApp
  4. Mark to make changes. 
  1. Hosting the site on Sites?
  1. Maybe using Google, let's see if we can get it on our domain.
  1. Yes, we would set site.icu-project.org or something, and the root page could redirect there.
  2. Review next week.
  1. TimeZone memory requirements, time zone formatting/parsing performance issues (Peter, Yoshito)
  1. >100K for CanonicalMap, OlsonToMeta, MetaToOlson; ticket #6645
  1. First try to optimize internal data structures, e.g., try using UChar* instead of UnicodeString objects
  1. ticket #6426 - large trie built for timezone parsing
  2. Currently putting all timezone names for locale into one big trie, including all variants
  3. A year ago, Mark wrote down parsing logic that did not require such a large trie
  4. Already smaller, 6414 (was 1.4M, now 180K)?
  1. should be done on parse, not when formatter created
  2. it is per language+territory+etc
  3. many users may only use a small number of zones
  1. have language-neutral data split out
  2. Will come up with a proposal and separate call
  3. Separate discussion group on this: Yoshito, Peter, Andy, Dave, Mark
  1. Yoshito will schedule a meeting: Dec 9, 2 PM PT
  2. Everyone should investigate approaches to improve this


  1. 2008/11/26


  1. 2008/11/19

  1. Available keyword values for a locale in a service(getKeywordValuesForLocale) (Kedar)
  1. Mark> For Java, we should probably have an Interface to encompass these APIs (we'll have to look back at your Java proposal).
  2. Steven> Hmm, commonlyUsed may be 'preferred' vs. 'available' as per CLDR.
  3. User wants:
  1. preferred list - being added to cldr.  An ordered list of what is preferred.
  2. commonly used / customary
  3. "what makes a difference"  - 
    en@collation=phonebook == en@collation=dictionary,
    de@collation=phonebook ≠ de@collation=dictionary
  4. "what is actually available/supported" - chinese calendar in russian may have english data
  1. What kinds
  1. calendar
  2. currency
  3. collator
  4. timezone?
  1. We can merge 'preferred' with 'customary' - only list customary calendars in the preferred list. ( a 'secondary' list could list non-customary calendar order? )
  2. 'what is supported' is hard to answer, may or may not need display. Don't do.
  3. 'what makes a difference' can be done by looking at the fallback chain.
  4. Potentially two lists:  'customary, in preferred order', and 'what makes a difference'
  1. customary, in preferred order
  1. currency: only currently active currency
  1. what makes a difference
  1. [all currency]
  2. [all calendars]
  1. verify that each service has an API for, and (CLDR) data is available for, and the documentation is correct for:
  1. all available keywords
  2. and all available values.
  3. customary in preferred order, verify that 'default' is the first item. 
  1. calendar: ok
  2. collator: no data in cldr
  3. currency : verify that we have this list
  4. timezone :
  1. Issue: what if this is a language only-locale?  i.e. preferred currency for 'en'. We will apply addLikelySubtags.
  1. Document that this is 'best available' data.  Default in the absence of other data. Different applications may have different preference orders.
  2. Also, '_DE' may expand to 'de_DE', including phonebook.
  1. OpenJDK Locale Enhancement project (yoshito,doug)
  1. http://www.openjdk.org/projects/locale-enhancement/ 
  2. Mailing list - locale-enhancement-dev -> http://mail.openjdk.java.net/mailman/listinfo/locale-enhancement-dev
  1. ICU resource bundle issues (yoshito)- re: "[icu-core] ICU resource bundle issues 11/5/08"
  1. Some design oriented issues
  1. Multi-level fallback (getByKeyWithFallback)
  2. ICU4J UResourceBundle as a Java ResourceBundle subclass?
  3. Available keys in a table
  1. Yoshito's "rough" proposal for ICU4J
  1. Separating low level data access APIs from user APIs.
  2. Deprecate UResourceBundle
  3. Create a new ResourceBundle class, but not as a subclass of java.util.ResourceBundle
  4. Create a new java.util.ResourceBundle subclass - ResourceCollection. which only support key-value mapping for a single bundle.  No locale inheritance is considered in this class.
  5. new ICU ResourceBundle#getBundle will initialize multiple ResourceCollection (each one is corresponding to a .res file) and build a locale inheritance chain.
  6. The new ICU ResourceBundle give you a view of logical resource tree (resources from multiple .res are merged)
  7. In the new ICU ResourceBundle, all get methods act like [currently internal] getByKeyWithFallback (fallback is always involved / can specify hierarchical key)
  8. ResourceCollection will parse the contents of .res and build the run-time friendly structure.  The raw byte[] contents will be released.
  1. Discuss Multi Level Fallback:
  1.  Original design: Java/Windows.  get 'Dialog Box' as a single structured item, with nested sub items. Would not want it to partially fall back.
  2. 'array' does not fall back- but a dialog may be key/value pairs
  3. don't want to break this behavior. can add API that will do this fallback, and hierarchical key [ get("calendars/gregorian/...")  ]
  4. could have a syntax to handle this:   dateFormatSymbols/4/…
  5. CLDR has some items which are ordered, such as collator contents.
  1. Discuss ICU4J UResourceBundle vs Java ResourceBundle  and proposal
  1. ResourceCollection (separate obj for [root+sr+sr_Latn]
  1. if there are aliases that point outside of your 'parent chain', does complicate things.
  2. should have all context in the bundle.
  1. currently evaluating prototype. 
  2. 'unflatten all at load time'  - #8 somewhat independent.
  1. historically, didn't do it due to concern about (1) load time, and (2) small % of items that are typically used. However, no measurement was done.
  1. Yoshito to write up more concrete proposal, motivation, and example
  1. make google doc, link here, link to bug(s)
  2. srl to look at C API issues.
  1. Q:  We have 'createXXXX(ResourceBundle)' and 'createXXXX(UResourceBundle)' - will we need new overloads?
  1. very few public APIs..
  1. Q: desync between C and J ?
  1. talk about corresponding changes
  2. the J ResourceBundle is a new API
  1. Q: we still want it to be a ResourceBundle?  Yoshito to answer.. design issue.
  1. most of these changes should be internal.
  1. Q: could document that keys containing slash in getByKey() do multilevel lookup and fallback
  1. already have lots of get*byKey. 
  2. slash is already a prohibited key name
  1. ex> supplemental data has:            "meta:Australia_Central_001"{"Australia/Adelaide"}
  1. drawback: unexpected behavior with addition of a slash,  "key/"
  2. look at the API.  top level vs. sublevel fallback.
  1. Bugs !


  1. 2008/11/12 —   Back to 11am PT meeting time! 

  1. Locale Canonicalization (Steven?)
  1.  6593 - bug
  2. hard coded mappings are in C and J:  IANA/BCP47, .NET, ICU compatibility, …
  3. icu/source/data/*.xml contain compat mapping that generate alias locales (de__PHONEBOOK, etc)
  4. Steven to file CLDR bug
  1. CLDR / tool status? (John)
  1. RBNF (John)
  1. Proposal to change the structure in the resource bundles keeping the RBNF syntax the same
  2. en { // current
         SpelloutRules {
            "%simplified:\n"
                "-x: minus >>;\n"
            "%default:\n"
                "-x: minus >>;\n"
                "x.x: << point >>;\n"
        }
        OrdinalRules {
            "%main:\n"
                "=#,##0==%%abbrev=;\n"
            "%%abbrev:\n"
        }
    }
  3. en { // proposed
         RBNFRules {
            "%spellout:\n",
                "-x: minus >>;\n",
            "%default:\n",
                "-x: minus >>;\n",
                "x.x: << point >>;\n",
            "%ordinal:\n"
                "=#,##0==%%abbrev=;\n"
            "%%abbrev:\n"
    }
  4. en {
             RBNFRules {
                  simplified {
                      "-x: minus >>;\n"
                  }
                  default {
                      "-x: minus >>;\n".
                      "x.x: << point >>;\n"
                  }
                  ordinal {
                      "=#,##0==%%abbrev=;\n"
                  "%%abbrev:\n"
                  }
            }
    }
  5. Rule Structure flattening (srl)
  1. old:
  1. Ordinal
  1. default
  2. x
  3. y
  4. z
  1. Spellout
  1. default
  2. x
  3. y
  4. z
  1. new
  1. ordinal
  2. ordinal_x
  3. ordinal_y
  4. ordinal_z
  5. spellout
  6. spellout_x
  7. spellout_y
  8. spellout_z
  9. checkwriting
  10. …
  1. Proposal is http://www.unicode.org/cldr/data/dropbox/misc/RBNFRules.html
  2. Concern with flattening (Doug)
  1. predefined sets
  2. behavior concerning using the last public ruleset
  3. could remove restriction on fixed set of keys
  4. multiple (private?) variations per key: ex, hebrew. 
  5. reimplement before defining CLDR data
  1. Difficulty: how to name sets across locales.
  2. Fallback from ordinal_foobar to ordinal
  3. John TODO: Comfortable with #4 above. Let's see how current rules would map to this format. 
  4. John to set up separate meeting to review RBNF design.
  1. ICU resource bundle issues (yoshito)
  1. Multi-level fallback (getByKeyWithFallback)
  2. ICU4J UResourceBundle as a Java ResourceBundle subclass?
  3. Available keys in a table
  4. Numeric index access to table items
  5. Sent a message to icu-core Nov 5th, please read for discussion 11/19.
  1. Available keyword values for a locale in a service(getKeywordValuesForLocale)
  1. Mark> For Java, we should probably have an Interface to encompass these APIs (we'll have to look back at your Java proposal).
  2. Steven> Hmm, commonlyUsed may be 'preferred' vs. 'available' as per CLDR.
  1. Bugs !



  1. 2008/11/05

  1. Meeting time? (Mark)
  1. Agreement: move meeting back to 11am PT
  1. LayoutEngine isBogus() (Eric/Jason)
  1. isBogus is new (not in released ICU)
  2. preferred model is to use the layout engine factory- which has always (?) had an error code.
  3. Public c'tor is already @internal: (LayoutEngine in 4.0) - users shouldn't be calling it directly
  4. Add error code to c'tor and all subclasses, change factory to use the version with the error code, remove isBogus
  1. Yoshito: add #6612 (ICU4J charset buffer bug) to 4.0.1 maintenance candidate list
  2. incoming bugs
  1. 6613 - srl to fix, peter to file new rfe
  2. 6599 - doug holiday
  3. 6603 - part of bcp functionality. Dave to document issues. CC: yoshito. M4.2
  4. 6609 - toPattern()/toLocalizedPattern(). yoshito to reply with comments and wontfix.
  5. 6611 - doug to look at content,  srl to check with IBM legal, possibly put an implicit contribution note on the bug submission form
  6. 6614 - peter to do.  4.1.2
  7. 6615 - peter to do.  cc: srl. 4.1.2 TRADITIONAL / GREGORIAN . UCalendarType
  1. UCAL_TRADITIONAL = UCAL_DEFAULT, no change (current behavior)
  2. UCAL_GREGORIAN: internally  calls setKeywordValue("calendar", "gregorian") - > @calendar=gregorian
  3. "ja_JP", UCAL_TRADITIONAL -> gregorian.  "no change"
  1. ICU Resource Bundle issues (Yoshito) - discuss next week.
  2. configure / ^M woes (Steven/Markus)
  1. if you touch configure/.in let srl know..
  2. autoconf is going to remove the bare ^M in a future version.
  1. getDisplayName bug in 4.0 - 6597 - errata page on 4.0?  (Steven/Yoshito)
  1. The fix was merged back to maint-4-0 - http://bugs.icu-project.org/trac/changeset/24940 
  2. ICU 4.0 download page was updated - http://source.icu-project.org/download/4.0.html#errata-2 
  3. Yoshito sent a message about the issue to icu-support/announce/design list on Nov 7, 2008

  1. 2008/10/29
  1. Broken Currency choice formats in ICU 4.0.  The problem is bad CLDR data.  Affects az, de, fil, fr, ko, th, vi, zh_Hant.  Mostly INR.  Also something funny with hu, bad in ICU, ok in CLDR.  (Andy, Doug)
  1. TZ bugs (Yoshito)
  1. 6597 - C GetDisplayName - done in trunk
  2. 6602 (cldrbug:1962)  new zone in 2008i - spec change in CLDR, need code change in ICU. Yoshito to file and number. John to file CLDR bug.
  1. New Google ICU committers: Fredrik Roubert [roubert] (Trondheim) & Shaopeng Jia [shaopengjia] (Zürich)
  2. Incoming Bugs
  1. Steven to file bug about testing with fake "current time", continuous monster test for dates. Done.

  1. 2008/10/22
  1. Reviews- thanks, more to do (see below) (srl)
  2. Proposal for encoding conversion security issues TR (Peter)
  1. would be good to have ICU samples in a separate ICU doc that refers to the Unicode TR
  2. Unicode TR will just have pseudocode
  3. Peter will make a proposal at next UTC meeting
  1. http://source.icu-project.org/repos/icu/data/trunk/meta/ - ICU Metadata project (Steven)
  1. Information about versions of ICU and their release dates
  2. Who has tested which version of ICU on which platforms
  3. Potentially auto-generate pieces of the download page
  4. ->  http://icu-project.org/xml/  and http://icu-project.org/dtd/
  1. Incoming Bugs


  1. 2008/10/15

  1. Converters toUnicode handling of illegal sequences, #5691 (Markus, Jungshik, Erik)
  1. Code point ranges: start+limit or start+end (see email) (Markus)
  1. Bug Cleanup, 4.1.1 (srl)
  1. 4.1.1, but no owner -> #6447 TODO: Google to figure out priority & scheduling & owner
  2. 4.1.1, very old code, but no reviewer -> #5503 , #5869
  3. 4.1.1, open and no reviewer: [query] - Please either assign reviewer or move to 4.1.2
  4. 4.1.1, has a reviewer: [query] - please complete reviews!
  5. Unscheduled, Active tickets with code and no reviewer: [query]
  6. 4.0 bug reviews-  please complete review or assign another reviewer! [query]
  1. Incoming Bugs


  1. 2008/10/8

  1. [moved to next meeting] Bug Cleanup, 4.1.1
  1. 4.1.1, but no owner -> #6447
  2. 4.1.1, very old code, but no reviewer -> #5503 , #5869
  3. 4.1.1, open and no reviewer: [query] - Please either assign reviewer or move to 4.1.2
  4. 4.1.1, has a reviewer: [query] - please complete reviews!
  5. Unscheduled, Active tickets with code and no reviewer: [query]
  6. 4.0 bug reviews-  please complete review or assign another reviewer! [query]
  1. tzdata 2008g
  1. ticket #6577
  2. for now- works through 2038, but need to add 64 bit (pre 1901, post 2038) - #5624
  3. done for C, not J yet-
  1. 64bit libraries in C (srl) - #6564 
  1. proposal: deprecate default behavior, replace with --with-library-bits=32/64
  1. Warnings on by default in C (srl)
  1. shadowing of index()  -Wshadow -?
  1. index is not one of the parameter names recommended by coding guidelines
  1. Umbrella tickets listed under milestone:4.2  - such as code cleanup for C
  2. ERA: #6547 - #define ERA,  langinfo.h selector for nl_langinfo()
  1. Incoming Bugs
  2. ustdio.h still has functions with "@draft 3.0" - should they become stable? (Peter)
  1. yes, @stable 3.0 - Peter to file bug → #6578

  1. 2008/10/1
  1. Data Thoughts:  would it be reasonable to make available with ICU releases a copy of the CLDR data that exactly corresponds to the included ICU data?  We at Google would find this convenient.  (andy)
  1. Time Zone Data Updates.  Do I assume correctly, in the case of metazoneinfo changes, that the ICU files will be the first to be fixed, and the CLDR xml files will follow?  (andy)
  1. autoconf problems. configure.in prereqs version 2.63, which was released by the GNU people 9 Sept 2008.  Suffice it to say that NO system distributions yet include this version, it must be built from sources. Do we really need this version? (I was trying to switch the compiler options to enable strict warnings.) (andy)
  1. code reviews (srl): Google tardy reviewing...
  2. 4.1.1 release - milestone-4-1-1 (srl, yoshito)
  1. bugs Incoming Bugs
  2. Doug sent out JDK Locale Enhancement note to openjdk announce list, to propose a new Project.
  3. Q: should milestones change

                2008/9/24

  1. Discuss M1 pending changes, announce proposed svn revision number for C and J M1

  1. John to finish collation data relocation in ICU4J
  2. Fire full build/testing for C now.  Wait for the collation change above for J.
  1. Request for participating ICU PMC from Benedicto Franco Jr. (yahoo)
  1. PMC membership is not based on company.  Tex was representing yahoo before, but it does not mean we'll fill in the position automatically.
  2. We agreed to follow following steps
  1. See if Benedicto can actually contribute things to the project, such as
  1. Active involvement in ICU mailing lists - design suggestions, proposals, comments...
  2. Contribution of new features/improvements/bug fixes
  3. Statement from his management addressing his commitment to the ICU project (for example, work time which he can use for the project) may help.
  1. After we confirm his contributions, we'll give him a committer access to the ICU repository.
  2. Then, we will vote for his involvement in the PMC
  1. Milestone BRS task assignment
  1. Steven/Yoshito will work on most of items.
  2. May ask someone's help on some tasks.
  3. If someone is willing to volunteer on some tasks, send a message to Steven/Yoshito
  1. Quick question about adding RuleBasedBreakIterator constructor that does not adopt data (Peter): icu-design mail 9/18, ticket #5858
  1. Agreed to add the option.  enum maybe used instead of boolean
  1. Resource bundle's behavior when fallback and alias to /LOCALE is involved.
  1. Yoshito posted the design question in the core ML on 9/24
  2. Yoshito to come up with more concrete examples and discuss this again in the next PMC.
  1. bugs Incoming Bugs

2008/9/17 (change to new meeting time 10:00 AM PT)

  1. New ICU tools work (Brian)

  1. 6353 - LDML2ICUConverter output .res directly ( minus Collator binary, some transliterator files )
  2. 4966 - ICU2LDML
  1. 4.0 cleanup
  1. http://bugs.icu-project.org/trac/ticket/5696 (Assigned to: markus)
    http://bugs.icu-project.org/trac/ticket/5415 (Assigned to: mark)
  2. 25 total remaining reviews for 4.0 stream: http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&groupdesc=1&group=revw&milestone=4.0&milestone=3.9.1&milestone=3.9.2&milestone=3.9.3&order=priority
  1. bugs Incoming Bugs

2008/9/12 (10:30 AM PT) - meeting moved from 9/10 due to IUC

  1. bugs Incoming Bugs

  2. Java Locale Update (Yoshito)
  1. proposed enhancements to JDK- so Java supports BCP47
  2. suggestion: start OpenJDK Project (under Internationalization Group)
  3.   Doc URL: http://docs.google.com/Doc?id=d3q3z35_10g9wwphg6
  4. want to put some implementation into ICU, as ICU will need to depend on similar functionality
  1. M1 small BRS (Oct 1)
  1. agreement was: make milestones more stable/usable.
  2. http://icu-project.org/processes/release.html 
  3. some items are a mess at the end, so do them at each point
  1. scan for copyright
  2. no-data test -
  1. (check srl's crash with custom data from iuc32)
  2. do, and tell people who worked on the code to be more careful
  1. bug reviews
  2. environment test
  3. uconfig test
  4. compile warnings
  1. andy to look at increasing the base warning level
  2. dave to look at pragmatics of turning warning level down on windows header
  1. send announcement
  1. tz - update next week
  2. version # in src and data - already done
  3. update configure
  4. check for non ascii
  5. static code analysis
  6. code coverage - just generate report, dont' hold release for it.
  7. code review - have them in progress
  8. API docs - generate, post as interim 4.1 docs, post errors as informative
  9. tag - ask people not to do destructive changes a week before.
  1. already the policy to put destructive changes on private branches
  1. urename.h
  2. API change report -generate and send to core.
  3. send a note to the list that the release is tagged.
  4. run exhaustive tests
  1. NO
  1. status comments
  2. compare icu4j/jdk
  3. update license files
  4. verify memory allocation
  5. eclipse plugin
  6. testmap
  7. udraft.h
  8. XLIFF
  9. sample/demo
  10. data portability
  11. locale explorer
  12. readme
  13. userguide
  14. download page
  15. public site
  16. big announce release
  17. update trac
  18. post-release cleanup
  19. online information update
  1. check-in date - Sept 24th. Discuss in PMC 
  1. if there's an error: we will discuss, fix major things, …  will create a branch only if needed
  2. minimum: exhaustive tests pass on reference platforms
  3. fix what we can, document the rest
  1. Codepage Detection
  1. Google: needs better detection with short fragments of text such as queries.  currently needs about 100-200 bytes. Discuss offline. Perhaps keep the entire histogram, may need weighting.
  2. Eric is getting more queries about this.
  1. cross compilation
  1. icu 4.0 source (re)packaging - Steven to update BRS document.
  1. Security notice:
  1. ICU4C 4.0 resolves the vulnerabilities CVE-2007-4770 and CVE-2007-4771 which were found in earlier versions of ICU. The standard ICU tests verify that these have been corrected, however, the updated versions of the previous tests may be run by applying the following patch to ICU 4.0: r24324. As well, ICU4C and ICU4J 4.0 resolve the issue underlying CVE-2008-1036. [#6175, #6198]
  2. Steven to Update 4.0 download page to link to 6175/6198 and mention 2022, HZ

2008/9/3 Agenda

  1. Java 5 (Yoshito)

  1. sent notes to core -
  2. large subset of ICU4J is used by CDC
  3. Option 2 (subset) not viable
  4. Option 3: Keep 1.3 subset
  1. Preprocessor
  2. Important features for ICU:
  1. Generics
  2. Autoboxing
  3. varargs
  1. Unimportant:
  1. for loop
  2. enums
  1. Option 1: Fork
  1. Fork become maintenance mode
  1. Proposal: (option 4- default implementation Java 5) - 'continue current strategy'
  1. Add new APIs under //#ifdef
  1. does cause maintenance burden
  2. can remove the old-support once eclipse moves forward
  3. Mark volunteered to propose some Java 5 API change to ICU as test case
  1. Do not use some features (for:, enums would be difficult to map inside an ifdef)
  1. Yoshito will come up with proposal, 2wks
  1. (Q:JDK Locale integration?: A: continuing investigation)
  1. Meeting on Sept 11th instead of 10th?
  1. Thursday Sept 11th or Fri Sept 12
  2. editorial: 10:30-6 Thursday
  3. friday at google?
  1. 10:30-11:30 PT meeting, then lunch?
  2. count:   5 IBM, 
  3. Andy to send out directions
  1. bugs Incoming Bugs

2008/VIII/27 Agenda

  1. JDK 1.5 (Yoshito)

  2. J2ME still continuing, based on CDC FP 1.1, JSR219, JDK1.4
  1. class library issue known, but issue is whether more recent syntax/bytecode can be introduced
  2. specify bytecode 1.5, cannot be compatible with prior jvm
  3. Mark: option 1, code fork. >1.4 adds robustness to code and interfaces.
  4. already have conditional compilation
  5. not just functionality, but inheritance
  6. code fork means data needs to be maintained
  1. cldr <-> icu version association
  2. need to provide old data, or limit data that moves
  1. next eclipse - ICU 4.0, june 2009
  2. Yoshito: option 2, provide a small subset to meet CDC requirements, and then have a full ICU4J (4.2+, Java 5+, etc)
  1. what APIs/implementation are affected? Yoshito to determine. And the transitive closure of APIs that those apis call.
  2. subset would be a 1.3-compatible part of ICU trunk- no back port needed.
  3. then, determine if this is feasable.
  1. Mark on rotation in Zürich (Mark)
  1. move meeting to earlier? 10:00a PT?
  2. Effective Sept 17th. Steven to send out calendar invitations.
  3. No meeting Sept 10th, maybe 11th
  1. ICU Release Browser (Steven)
  1. http://apps.icu-project.org/icu4jweb/icuinfo.jsp
  1. http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uproj=icu4c
  2. http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uproj=icu4c&uvers=4.0 
  3. searchable by attribute: http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uwhat=match&uattr=os%2Bversion&uval=AIX+5.3 
  1. http://icu-project.org/xml/icuinfo.xml 
  1. IBM specific testing: http://icu-project.org/xml/testing_ibm.xml
  1. phasing out 'reference' in public docs
  2. regularly vs. often doesn't mean anything
  3. regularly here means 'a binary is available'
  4. could add a 'distribution format'
  5. Q: 'if i pick up the source, did it pass?'
  6. Binary available, is us
  7. What is useful information
  1. binary available
  2. did tests pass on this platform ( yes- completely, unconfirmed [no reason to list this], known failure )
  1. link: 'if you are interested in helping to fix this platform'
  1. 'tested by: IBM'
  1. 4.2 reference platforms platform decisions?
  1. bugs Incoming Bugs
  2. Collation Sensitive Regex (Eric)

2008/8/20 Agenda

  1. Update ICU to require JDK 1.5, eg Generics in APIs.

  1. Eclipse team's note
  1. JFace (which uses ICU4J) currently targets CDC Foundation 1.0, and as far as I am concerned, should not be moved to 5.0 unless there is a compelling reason for doing so.  Se we would very likely keep using the particular subset of ICU4J we are using now, so that it should run with whatever version of ICU4J someone decides to bundle with their Eclipse-based product.  What we ship with the SDK is an important question of course.

    Also, to understand this a little bit better - as far as I know, there are parts of ICU4J that need to be updated regularly, as new currencies, time zones, and languages are introduced, correct?  If ICU4J decided to use Java 1.5 for upcoming releases, would it be possible to do these kinds of updates in a way that they benefit both the newest version of ICU4J and an old maintenance branch?  I guess I am asking if branching is really going to be that big of an overhead.
  2. This would push everyone except Equinox and SWT up to requiring Java 5. Since they are talking about after the 3.5 release, this would impact 3.6/4.0.  There was a lot of discussion about this in the context of e4 but nothing conclusive.
  3. No consensus, so it probably is worth bringing up in the Arch call.  I've included some of the responses I got to my query.

    Equinox doesn't seem to be too effected by the ICU plugin, but JFace is and probably wouldn't move up the the 5.0 EE.  Is it that people want to use 5.0 in the ICU API or that they want to use the 5.0 language features (like enum or foreach loops) within the compiled code?

    The Arch call is probably where they would have to think about what Eclipse itself will do.
  1. Eclipse platform's usage
  1. http://wiki.eclipse.org/ICU4J_Usage_in_Eclipse
  2. http://wiki.eclipse.org/index.php/ICU4J 
  1. Generic issues
  1. Sun Java SE 1.3/1.4 is out of support
  2. J2ME - http://java.sun.com/javame/reference/apis.jsp 
  1. Connected Device Configuration - Foundation Profile 1.1 (JSR 219) is based on JDK 1.4
  1. javac in JDK1.5 cannot produce Java bytecode works on JRE1.4 or older with source=1.5
  1. retroweaver can transform 1.5 bytecode to 1.4 - would not solve class library issue.
  1. Solutions
  1. (option 1) code fork
  1. code is effectively forked with 4.0 maint stream.
  2. issue: maintaining data going forward
  1. could only backport critical data changes.
  1. (option 2) separate "core" subset of ICU4J work on CDC1.0
  2. (option 3) keep Java 1.3 syntax
  1. revisit next week…
  1. ICU4J class file issue (2.6.1) - http://bugs.icu-project.org/trac/ticket/6505 
  1. Eclipse bugzilla - https://bugs.eclipse.org/bugs/show_bug.cgi?id=216147
  2. ICU meta file: http://icu-project.org/xml/icumeta.xml
  1. IUC32 status (Michael)
  1. Final slides due 8/25 now, send a note to Kevin if you did not receive any comments.
  2. Eric, Yoshito will be at IUC- no PMC at regular date/time Sept 10th
  1. move to thursday (Sept 11) or friday (Sept 12)?
  1. bugs Incoming Bugs

2008/8/13 Agenda

  1. bugs Incoming Bugs

2008/8/6 Agenda

  1. C trunk broken-?

  1. Need to swat - replace primitive types in C according to coding guidelines.
  1. yyyy, will add to CLDR, FlexibleDateTime stuff needs some fixes.
  2. C:  version updated to 4.1.1
  1. J already updated - data version to be updated - Yoshito to update today.
  2. C: translit r/t timebombs removed, bug filed in CLDR
  3. Agreed; remove old timebombs where possible. Yoshito to add to BRS
  1. review short bugs (monthly)
  2. Small design items in email
  1. Raising methods into Collator. Will have NOOP implementation; documentation will indicate that subclasses should support but don't need to; these are requests. Agreed that this should be done more generally; Mark to file concrete API changes on design list.
  2. PunyCode public. Mark to add Transform interface.
  1. Website:
  1. email: Proposed Demo page additions - Mark, Andy to do.
  1. Update ICU to require JDK 1.5, eg Generics in APIs. Eg use generics in APIs and implementations, allow us to have parallel APIs to JDK (right now we have holes). Options?
  1. Maintain 4.0 for Eclipse, split if necessary
  2. Have core that is unchanged (Mark's comment - doesn't meet requirements)
  1. Bugs


2008/7/30 Agenda

  1. Schedule for Ticket #6426, excessive memory use in TimeZone Formatter (Peter)
  1. Apple needs by 9/30 (and needs to roll back to our 4.0 sources), can do some or all of the work. Discuss possible approaches?
  2. Actually, Markus already checked some fixes into trunk that reduce the size of the parsing data structure and go most of the way to addressing this.: That is Ticket #6414. That is probably enough to address the Apple issue, so that #6426 can be for more long-term solutions such as lazy initialization.
  3. Andy - another related issue, resources not flushed from the resource bundle cache in ICU4C. He is working on some code to address this. He will create a separate ticket for this, it is a general issue.
  4. http://bugs.icu-project.org/trac/ticket/6414 should improve the memory footprint.
  1. Performance tests/metrics for new features (Peter)
  1. Some way to instrument tests to measure peak memory usage during test? Needs investigation
  2. More testing of memory usage of development releases of ICU
  3. Peter will create a ticket
  1. https://bugs.webkit.org/show_bug.cgi?id=15914
  1. Implement Unicode functionality using GLib.
    This could be a fairly easy way to shave off 10M shared library memory use due to ICU...
  1. TimeZone#setDefault problem
  1. Yoshito sent the problem description to the core ML.
  2. If no objections, Yoshito will make the change as described in the message.
  1. Bugs Incoming Bugs

2008/7/23 Agenda

  1. trivial bugs (continued from last time)

  1. from last week - some bugs are trivial, take only a little bit of time to do
  2. we'll have spent more time looking at the bug than fixing it
  3. "fix-it day", to just fix bugs
  1. reminder message
  2. bugs by owner, 4.1.1 
  3. Bugs by owner, 4.2
  1. need a way to have these bugs show up as higher priority, especially earlier in the release.
  2. These are (trivial || minor && low weeks) bugs.
  3. Tactical: Mark all bugs which are quick to do, and then review them.
  4. Review once a month, at the start of the month
  1. ICU functional area contacts
  1. ICU Functional Area Contacts 
  1. Go over these, make sure appropriate
  1. Aligned to the current trac component keywords - are the current keywords appropriate?
  2. ICU 4.0 components assignment - for comparison
  3. split up formatting?
  1. numbers (currency, ...)
  2. dates (+timezone)
  3. "other"?
  1. Ticket Priorities and Dispositions
  1. Priorities:
    blocker — Item prevents shipment of ICU and/or other major tasks to proceed.
    critical — Major crashing bug or a feature whose absence prevents usability.
    major — Important feature or bug fix.
    +medium — Bug of medium importance.
    minor — An item which would be good to do, but isn't very important. Do if there is time.
    trivial — An item of very little importance and impact, such as a minor typo in documentation.
    zero — An item which is valid, but not currently considered important enough to fix or do. These items may be left open as zero instead of closed as 'wontfix' so they are searchable in case their priority becomes important later.
    assess — The assignee of this item is requested to determine what the bug priority is, if any. It may be that the item is not valid or should not be done.
  2. Dispositions:
    Tickets may be closed with the following dispositions:
    fixed — Only the person listed as "reviewer" for this bug should close the bug as fixed, once they have reviewed the code.
    invalid — The ticket is not valid, it is not a bug or feature request. It may be spam, or just useless text.
    wontfix — This will not be fixed (or implemented) by ICU.
    duplicate — The bug has already been filed.  Add a comment to the duplicate bug saying, "This bug is a duplicate of #1234".     If the duplicate adds anything to the discussion, add a comment to the remaining bug such as, "Bug #4321 was closed as a duplicate, but it contains some interesting test cases."
    worksforme — The bug was not able to be reproduced.
  3. Issues
  1. Gap between major and minor - lots of 'medium' category.
  2. worksforme conflates  "this was broken before but now fixed",  "could not reproduce this, even in an earlier version"
  1. "moot" - "does not matter anymore" (could refer to functionality that is no longer supported)
  2. Must close worksforme with a comment, "works in C4.8" or "works fine in r43218"
  3. could add a field: "works in r12345" or "fixed in 4.8"
  4. Could add a button, "which releases have this code"
  1. action items- DONE.
  1. remove blocker
  2. add medium
  3. drop trivial
  1. 4.2 planning
  1. Review each company's load and move some tickets across companies.
  1. example: may have Google:100, but under Yoshito. If IBM has no interest, transfer to Google person.
  2. Query: Bugs with a Load set and no reviewer
  1. bugs Incoming Bugs

2008/7/16 Agenda

  1. tzdata2008d

  1. Uploaded the files to the ICU project server (zoneinfo and metazoneInfo)
  2. Updated ICU4J Time Zone Update Utility to support metazoneInfo.
  3. supplementalData which contains zone aliases and region mapping data is not included, because the single version does not work well with all ICU versions.  We agreed this in the last PMC meeting.  supplementalData was updated along with zoneinfo/metazoneInfo in ICU4C/J SVN trunk only.
  4. Apple has a build system problems with the CLDR dependency - metazoneInfo (and supplementalData), see next item
  1. Need to be able to update Olson data - and now metazoneInfo - independently of ICU (Apple)
  1. In Mac OS X, the Olson data is used directly by many projects: ICU, BSD, etc.
  2. All of those projects need to be in sync, using the same version of Olson data. We need to be able to update the Olson source data and and all dependent source data (which now includes the metazone data) independently of ICU sources. We have a TimeZoneData project that contains Olson data and is submitted independently of ICU.
  3. We cannot use zoneinfo.res and metazoneInfo.res. The Apple build process requires that we submit sources, and moreover, the other clients of the time zone data do not understand zoneinfo.res.
  4. The ICU build process does permit using out-of-source-tree Olson data to override the data in the ICU source tree. If the makefile variable TZDATA points at tzdatayyyyv.tar.gz (a version of the Olson data), source/data/Makefile.in will override the zoneinfo.txt in source/data/misc. It does this by running the tools in source/tools/tzcode to take the Olson data pointed to by TZDATA, and producing a new version of zoneinfo.txt that gets used instead of the one checked into the source repository. In our case, TZDATA points at /usr/local/share/tz/tzdatayyyyv.tar.gz, which is created by the TimeZoneData project. So for example, in Mac OS X 10.5.4 we have the following in /usr/local/share/tz/:
  1. -rw-r--r--   1 root  wheel  190412 Oct  1  2007 tzcode2007h.tar.gz
  2. -rw-r--r--   1 root  wheel  162187 Oct  1  2007 tzdata2007h.tar.gz
  1. Although 10.5.4 is using ICU sources derived from ICU 3.6 (which had Olson 2006j), and those have been modified for Apple's Windows builds to have zoneinfo.txt with Olson 2007k, when built for Mac OS X, ICU will pull the Olson 2007h data from /usr/local/share/tz.
  2. What we need now is add metazone source data (txt or xml) to this process. We need to be able to put the relevant source files in the TImeZoneData project, and the ICU build needs to allow metazone source data in some place like /usr/local/share/tz/ to override the metazone info in the ICU sources.
  3. => Agreed in principle to makefile changes that allow metazoneInfo.txt* from outside the ICU source tree to override the one in the ICU source tree, Peter will file a bug and submit proposal to icu-design. (* or whatever metazone-related files fall out as a result of http://bugs.icu-project.org/trac/ticket/6444 ).
  1. Trac system usage issues
  1. Use of "accept" - ignore "accept" - used for nothing
  2. Priority for enhancements vs. implementation bugs
  3. How to handle trivial bugs
  4. Usage of Load - 100% means intention to complete the bug; doesn't mean fall-on-your-sword commitment. May have multiple orgs with 100%. 50% means willing to work on it and/or trade off working on that with other 50%.
  5. Add "reporter" to spreadsheet - Andy to do.
  6. Automatic ticket assignment - a new incoming bug will be reassigned to functional area expert's bucket automatically
  1. Yoshito's comment - I like to go through incoming bugs in weekly PMC meeting and assign each of them to developers at that time.  A new incoming ticket may be high priority for some companies, but could be low priority for other companies.  If a ticket is assigned to a developer who belongs to the latter group, people in the former group may miss it.  In major release planning phase, each company goes through the whole ticket list once (like we're doing now).  But once this is done, we rarely inspect all of open tickets.  I have no problem to create the functional area assignee list - when none of PMC member companies are interested in a new incoming ticket, the ticket may be moved to the functional area assignee.
  2. Shared doc - developer per each functional area -> Yoshito will create the google spreadsheet.
  1. String search?
  2. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/7/9 Agenda

  1. 4.2 schedule

  1. more realistic dates - Jan 1 is not a likely ship date
  2. decree a milestone when a significant feature is done
  3. milestone prevents us from drifting, and reduces catch-up at end (when too many items are deferred to the GA timeframe)
  1. do some BRS items  (LRS)
  1. use branches - keep the trunk more stable.
  2. treat the milestones as real deadlines - some lead time before release date, ~1 wk
  1. could use maint-4-1? or just don't put small fixes in.
  2. code slush to maint-4-1 a week before each milestone release, and then BRS
  1. 4.2 planning
  1. Policy for critical/major items
  1. unscheduled critical/major are being moved to assess..
  2. please either move to minor/trivial (unscheduled), or set a milestone 4.2 or 4.4
  3. Names: "trivial" sounds like an amount of work.  Is it a Severity?  Priority: ROI. Severity: Urgency+Fixability. (applies to bugs, not features.)
  4. trac mainpage links to: http://www.icu-project.org/processes/ticket-lifecycle.html 
  5. We may need more explanation for each keyword. The above page is a good place to document this.
  6. Steven and Yoshito to draft update with explanation for keywords
  1. 7/23 - 4.2 candidates review
  1. fill in Load:   IBM:50 Google:100 Apple:100 etc ( % of interest&&availability to work on the bug)
  1. tzdata 2008d status (Yoshito)
  1. metazone mappings change (otherwise, fallback mapping will occur)
  1. (olson data) -> zoneinfo.txt/.res
  2. (CLDR metazoneInfo.xml) -> metazoneInfo.txt/.res
  3. (CLDR supplementalData.xml) -> supplementalData.txt/.res (contains other unrelated things..)
  4. proposal1: include zoneInfo and metazoneInfo but not supplemental data.
  5. proposal2: separate data out so that timezone-only .res files are created. (so that only those are updated.)
  1. CLDR 1.6.1
  1. checked in
  1. Update instruction change
  1. zoneinfo.res, metazoneInfo.res, (skip supplementalData.res?)?
  1. ICU4J Time Zone Update Utility changes
  1. needs to be updated to install metazoneInfo.res
  1. hope to wrap up by end of week or early next
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody


2008/7/2 Agenda

  1. 4.0 - ready to ship? (steven)

  1. translit issue - give it 2 hours to investigate
  2. Draft 4.0 release page:  http://icu-project.org/download/4.0.html
  1. 4.1 - update version in trunk? (yoshito)
  1. may be timebomb failures
  1. 4.2 schedule
  1. From 6/11 meeting: "May 1 GA" - OK.
  2. Need intervening milestones
  3. CLDR: OK with CLDR release in the May timeframe.
  4. No Unicode update in timeframe. - 5.2 in fall?
  5. n milestones, 3mo apart - adjust based on interesting features
  1. M1 Oct 1, 2008
  2. M2 Jan 1, 2009
  3. M3 March 15 - DCUT API Freeze
  4. M4 Apr 1, 2009 
  5. M5 April 15 - code freeze
  1. GA May 1 2009
  1. Please clean up 4.0 tickets
  1. 40+ reviews
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
  1. in 2-3 weeks - go over 4.2 bugs, determine who is interested..
  1. 6397 yoshito
    6398 andy
    6399 xji
    6400 yoshito
    6408 doug
    6409 yoshito
    6364  srl
    6383 srl
    6389 doug
    6390 doug
    6394 andy
    6410 yoshito
  2. July 23rd - go over 4.2 bugs, determine who is interested in what, set 'load' field.
  1. "interest" here means "you are willing to work on it".
  1. Transliterator bug…  user error.
  1. \u3051\u30c0  -> Keダ  ( should be 'keda'
  1. works from a custom intltest test
  1. \u03cc  NFD should change this char
  2. Due to a bad custom rule named 'NFD'. Filed bug 6417..


Old 4.0 situation room ( moved on July 9th, 2008  by srl )

  1. john- cldr 1.6d03 (RC)
  1. #6378
  1. Integrate Data from CLDR release-1-6-d03
  1. data
  1. Current
  1. 4.0
  1. defect
  1. blocker
  1. 06/23/08
  1.     -
  1. C: r24307 ticket:6378: Integrate CLDR release-1-6-d03
  2. J: r24308
  1. xiaomei - interval (c /j DONE?)
  1. #6157
  1. C: r24280
  1. doug- plural
  1. J: #6391 r24290
  1. claire - plural
  1. C: #6374:  r24299, r24310
  1. dave- resource data
  1. comes automatically?  Yes.
  2. J: #6124 r24287
  3. C  #6149  r24285 typos in comments??
  1. srl- cygwin mingw build changes #6322 (possible), possible icu4j doc
  1. C #6098 ( version, release, etc )
  1. C r24291 update readme, add API change report.
  1. mark- crashing bug in translit. Mark wants to fix: 
  1. #4200
  1. Transliterator crashes on certain data
  1. transliterate

  1. 4.0
  1. defect
  1. critical
  1. 11/03/0



ICU 4.0 Situation Room release schedule

2008/6/25 Agenda

  1. Tex @ NetApp - http://icu-project.org/projectinfo.html 

  1. (I see you are on the document, good. -srl)
  1. 4.0 status (Steven/Yoshito)
  1. CLDR status (John)
  1. 1.6d02 - into icu 4.0d03
  2. some further changes..
  1. az
  2. weekend
  3. korean
  1. close..
  2. probably final @ CLDR meeting 7/1/08
  3. plural rule format change
  1. plurals don't have getavailablelocales ( or getfunctionalequivalent ) - doug to file api
  2. isolated change - plurals.txt
  1. 1.6d03? (RC?) - probably tomorrow 6/26
  1. Any tickets not yet merged into 4.0 release stream?please update w/ bug number - above list, situation room
  2. trunk is post-4.0 effectively,  notify icu-core (for srl and yoshito) with revision/ticket for merge into branches/maint/maint-4.0 which will be re-tagged -
  3. Code review - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=revw&order=priority&revw=%21
  4. All please check d03,  download page, readmes.
  5. icu4c 4.0d04 (RC) - EOB 6/27? 
  1. see 'situation room' above
  1. API question: PluralRules.select vs. PluralFormat.format (Yoshito)
  1. can change after 4.0 - @draft 3.8
  2. (long)1 vs (double)1.0
  3. future: need to look at  "1 hour" vs "1.0 hours"
  4. format (# of decimals) decoupled from primitive type (long vs. double)
  5. only on java
  6. bug #?
  7. Doug will take bug #
  8. format method is final
  9. send exact API changes to icu-design (NB: api docs will need to be regenerated)
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody


2008/6/18 Agenda

  1. Welcome Brian Rower (IBM)

  2. 4.0 status (Steven/Yoshito)
  1. Any remaining items to be included in 4.0?
  1. Please start cleaning up your 4.0 tickets - move them to 4.2/UNSCH if you have no plan to work on them for 4.0
  2. Any non-trivial tickets still working?  Please let Steven/Yoshito know about them.
  1. (you can list them here...)
  2. EBCDIC issues in progress (mostly test issues)
  1. late API changes are on icu-design - interval format parameter, UCharacter
  1. CLDR status update
  1. Dave has merged changes into trunk
  2. two upcoming merges:  one targetting the branch (today/tomorrow), and one later one.
  1. Reminder - creating branch the end of tomorrow (Thu June 19)
  1. will proceed with ICU d03 on June 23 as planned, but know there may be some cldr-related churn
  2. (CLDR: tag d01 at end of day after default content resolved?)
  1. Please work on code reviews!
  2. BRS items - any problems so far?
  1. environment tests (eric)
  1. timezones on mac - gmt offset not a multiple of :15 - saudi arabia, africa, ..
     - eric's note subj: '[icu-core] environmentTest Timezone Failures' 6/12/08 9:27 PM PT.  Could make this a warning in this or future release. Historical zones.
  2. ko_KR.CP949 locale / backslash / ucbuf  ..  default codepage handles "\" or "\u" incorrectly.  Fixed in most places, but difficult with included resource. "\uXXXX"  -> something like "₩uXXXX".
  1. recommendation: map 0x5c <-> U+005c
  2. action: verify that it is not a regression from 3.8.1. Document in readme. Investigate in 4.2.
  3. main() { puts(ucnv_getDefaultName()); } will return default codepage - see what is returned under ko_KR.CP949
  4. post 4.0:  prefer the default converter that DOES map 0x5C <-> U+005C
  1. test without data - done (claire)
  2. no namespace  - needs to be added to BRS (andy)
  3. uconfig test (srl) - usearch will require brkiter
  1. 4.0 download page
  1. Steven to update readme with a link to generate a list of bugs fixed
  1. CLDR default content wrap up

http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody2008/6/11 Agenda

  1. 4.0 Schedule (Steven)

  1. Branch - Scheduled for June 13th
  2. but CLDR not ready?
  3. Dave made changes to CLD
  4. Changes to LDML2ICUConverter tool for data differences
  5. Dave will be checking updated data into a branch, pull into icu4j
  1. plural rule on icu side waiting to be hooked up
  1. CLDR merge: Mon, Jun 16th - then some bug fixing
  2. Branch: end of Thu June 19th, and communicate on Fri, Jun 20th any issues -  Steven and Yoshito will merge as needed
  3. d03: Mon Jun 23
  1. 4.0 BRS
  1. http://www.icu-project.org/processes/release.html 
  2. http://spreadsheets.google.com/ccc?key=plRXpLeUpVd2kRTY-HJGEFg
  3. look at bugs with code and no reviewer http://bugs.icu-project.org/trac/report/18
  1. API Tweaks - Xiaomei interval format
  1. small, will send a list to design list
  1. 4.2 Schedule
  1. Steven's concern: March ship brings end of release at end of year
  2. Issue: CLDR schedule?  Gap with ICU vs CLDR brings unexpected difficulties
  3. CLDR schedule not decided
  1. longer candidate timeframe
  1. If the ICU M release is usable, will be some value
  2. at least 1 or 2 rigorous milestones - 
  1. end of March/beginning of April?
  2. more development on branches
  1. develop milestones in branch
  2. could have maint-4.1 
  1. keep the trunk clean - continuous integration?
  2. may want more policy: keep things in branches unless it is in shape, with test coverage
  3. Andy's Mark's Suggestion, for large things, review on branch before merge?
  1. IBM would like to see release in June?
  2. Tentative GA: May 1
  1. Withdraw spoof detection. #4790 (Andy)
  1. in a branch, headers checked in
  2. will withdraw headers for 4.0
  3. verify that net result is no change in trunk, and change bug to 4.2



(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
2008/6/04 Agenda

  1. Bug 6321, incorrect short time zone names.  Hot, I'm looking at it but, could use help.  Snarled in metazone code.  John?  (Andy)
  2. Encoding Selection (Andy)
  1. makes heavy use of STL - not in the form to go in now.. involves memory management
  2. pushed out to 4.2
  1. 4.0 Schedule (Steven) 
  1. http://www.icu-project.org/processes/release.html updated
  2. spreadsheet: status
  3. Will send out task items, to be assigned today and tomorrow
  1. CLDR 1.6 Schedule (Peter)
  1. CLDR 1.6 BRS will start June 10th
  2. John is already running tests
  3. Expect data to change until end of next week
  1. 4.2 tentative schedule (Steven)
  1. Slush: Feb 4, 2009
    DCUT: Mar 4, 2009
    GM: Apr 29, 2009
    GA: May 13, 2009
  1. CLDR Plan: open up Survey Tool right after 1.6
  1. need data firming in Indic
  1. Won't be a major Unicode release until later
  2. Mark: end of March? to avoid holidays. End of quarter is good



2008/5/28 Agenda

  1. Code reviewers for String Search: Mark & Andy? (Eric)

  2. We have an internal customer for ticket:2469 (Regexp, support for Unicode Regexp Level 2) and would like to see it scheduled for 4.2. (Eric)
  1. VS 32/64 bit -
  1. if the intermediate and output dirs are different, can have simultaneous build.  Eric to file bug based on support mail .
  1. 4.0d02 status (Steven, Yoshito)
  1. seems to be building clean on C after fixes
  2. J: errors?
  3. any pending changes?
  4. tag 3pm PT today?
  5. TimeUnitFormat - in J (6303), but won't make it in C for 4.0. Please move/refile bugs for 4.2
  1. 4.0 Release Items: Yoshito will post worksheet after this meeting.
  2. 4.0 reviews
  3. Copyright - Steven is checking on status


(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

 2008/5/21 Agenda

  1. 4.0d01 status (Steven, Yoshito)

  1. can check in code at the same time as design proposal
  2. ICU4J binary/source compatibility:
  1. nothing stated
  2. many people run against the prebuilt jars
  3. userguide should state the policy
  1. future: discuss clear statement on icu4j stability runtime vs. source.
  2. cutoff for DCUT tag- today, 3pm pacific - non functional stubs OK
  1. pending for #6291 - peter
  2. charset selection (design posted today)
  3. date intervals
  4. simple durations
  1. delay dcut 1 week  for charset selection/date intervals/simple durations?
  1. could still go out with d01 but not call it the final DCUT
  2. still get things in by 3pm if possible (especially changes to existing APIs)
  1. CLDR data generation status
  1. John and Dave to coordinate getting consistent results from LDML2ICUConverter
  1. Welcome Jason Spieth
  2. Transliterator timebomb status (4.0)
  1. Thai-Latin Instantiation seems to be fixed for C  (  ( no data test yet ) 
  1. Not the case for J... ?  - was already in J
  1. unicode r/t tests still were failing, moved to post 4.0 timebomb
  1. If this is not fixed by CLDR 1.6/ICU 4.0, then these tests should be removed, and moved into CLDR (post 4.0)
  1. date format pattern generator timebomb, J, C?
  1. Yoshito will send a note out
  1. API stability status update
  1. ICU4J (Yoshito)
  1. API Change report for ICU4J 4.0 -> http://icu-project.org/~yoshito/api40/icu4j_compare_381_400.html
  2. Charset - @draft ICU 3.6/3.8 are promoted to @stable except CharsetCallback
  3. BiDi / Plural format / Relative date format / Duration format / Time zone transiton / Rule based time zone - keep @draft ICU 3.8
  4. GlobalizationPreferences - keep @draft ICU 3.6
  5. UnicodeSet.XSymbolTable - keep @draft ICU 3.8
  6. Indian/Taiwan calendar - changed from @draft ICU 3.8 to @stable
  1. ICU4C (Steven)
  1. analysis not done
  2. will check in and announce changes for discussion today
  1. ICU 4.0 release tasks (Yoshito)
  1. will consolidate tasks, and make copy for each version
  2. discuss next PMC
  1. ICU4J Locale Service Provider for Java 6 (Yoshito)
  1. Separated root - icu4j/localespi
  2. Separated jar file - icu4j-localespi.jar



(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/5/14 Agenda

  1. IUC32 status?

  2. New locales for 4.0 ( e-mail from John )
    ii_CN - Yi (China)
    fr_SN - French (Senegal)
    ha_Latn_NG - Hausa (Latin, Nigeria)
    pa_Arab_PK - Punjabi (Arabic, Pakistan)
    zh_Hans_HK - Chinese ( Simplified, Hong Kong )
    zh_Hans_MO - Chinese ( Simplified, Macao )
    zu_ZA - Zulu ( South Africa )
    ne_NP - Nepali ( Nepal )
    ne_IN - Nepali ( India )

    No objections noted.  If anyone has any additions/deletions, please contact John.
  3. ICU 4.0 status
  1. API freeze coming up next week - May 21
  2. Will update version to 4.0 at DCUT (may 21 ,D01) as well as data version to '40'
  1. this affects timebombs- 5608, 4009 -
  1. CLDR Integration status
  1. John has snapshot with all tests working on the C side.  Will integrate to trunk within next day or 2, followed by building/integrating data for J.


(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
 2008/5/7 Agenda

  1. Locale/Language Fallback and Java Locale proposal (Yoshito, Mark)

  1. http://docs.google.com/Doc?docid=dfqr8rd5_66cmjvtgfn&hl=en (Mark's Locale/Language Fallback)
  2. http://docs.google.com/View?docid=d3q3z35_10g9wwphg6 (Yoshito's JDK locale enhancement proposal)
  3. bcp47
  1.   The relevant tickets are:
       http://bugs.icu-project.org/trac/ticket/6204   (4.0)
       http://bugs.icu-project.org/trac/ticket/6177 (4.0)
       http://bugs.icu-project.org/trac/ticket/5767 4646 support ( not scheduled )
  2. cldr bcp47 syntax: http://unicode.org/cldr/bugs/locale-bugs/docs?id=1730 
  1. LDML2ICUConverter CVS tag (Mark)
  2. ...
  3. Design Proposals - Dave - ticket 6228



2008/4/30 Agenda

  1. CLDR Data integration status (John)

  1. tool issues (LDML2ICUConverter, LocaleMap null) (Mark)
  2. change in supplemental / plural?
  1. plurality of supplemental parsers (Peter to look at, may be a few days before he can)
  1. John back in town tomorrow 5/1 - John to send out status of integration
  2. Dave investigating which codebase was 1-5-1, not tagged (1.5.1-RC was tagged)
  1. IUC 32 abstract submission status
  1. submitted…
  1. JDK Locale proposal (Yoshito)
  1. http://docs.google.com/View?docID=d3q3z35_7c2rcqwg4 (older version as of this writing, will be updated)
  2. Had some meetings to incorporate feedback
  3. plan is to send to openjdk i18n list sometime next week, more meeting(s) next week
  4. to give overview later in this meeting?
  1. 4.0 status (Yoshito, Steven)
  1. http://bugs.icu-project.org/trac/milestone/4.0
  2. 3 weeks until API freeze / D01 May 21st - at least API proposals and stub code. Keep this schedule.
  1. spoof detecton on branch (Andy?)
  2. likelysubtags on branch - waiting on LDML - API4C submitted to design, API4J to be submitted (Dave)
  3. Interval/Duration/Plural has API (Mark?)
  1. Please keep an eye on reviews and bug assignments.
  1. 4.0 Reference Platform (Steven, Yoshito)
  1. IBM's reference platform list
  1. Java
  1. AIX 6.1, IBM Java 5 (64-bit)
    Solaris 10, Sun Java 5 (32-bit)
    Windows Vista SP1, IBM Java 5 (32-bit)
  1. C/C++
  1. AIX 6.1, 64 bit VA 9
    Solaris 10 Sparc 64?, Sun Studio 12
    Red Hat Enterprise Linux 5, x86-32? GCC 4.1.2 (Feb 13, 2007)
    HP 11iv3 - aCC A.6.14
    Windows Vista SP1 32 bit, Visual Studio 2005
    Windows Server 2003 64 bit, Visual Studio 2005
  1. Apple builds & tests for the following:
  1. Darwin / Mac OS X 10.5, PowerPC 32/64 & x86 32/64, gcc 4.0.1
  2. Windows XP SP2, x86/32, Visual Studio 2005 (compiler & linker) + Cygwin (make)
  1. Q: Reference list  including platforms from other parties?
  1. IBM's list driven by IBM's reference support.  IBM-ism.
  2. reference platform list (for C) [historically] determines what platforms binaries are built for.
  1. We do build on other platforms
  2. http://source.icu-project.org/repos/icu/icu4j/tags/release-3-8-1/readme.html#PlatformDependencies
  3. http://source.icu-project.org/repos/icu/icu/tags/release-3-8-1/readme.html#HowToBuildSupported
  1. Public only cares about (a) which platforms are tested, and (b) which platforms are available for binary download.
  1. could have a page for people to report in which platforms build OK
  2. could semi-automate this with configure generating a 'report' to send in
  3. General Public does not care about reference vs. tested
  1. Proposal:
  1. public site: list of tested versions,  list of downloadable versions
  1. change 'reference' to 'regularly tested'
  2. list 'binaries are available' for the list where binaries are available
  1. Vista 32 binaries probably good for w2k, xp, 2003…
  1. internally: keep track of who tests which platforms
  1. which PMC members build which platforms regularly
  2. which PMC members build which platforms occasionally
  3. which external people claim platform X works on ICU  #.#
  4. which external people have claimed in the past ICU worked on some platform ('rarely tested')
  5. … combine these into one external list for the ICU web page
  1. Google's list of regularly tested platforms

ICU4C

  1. x86 Linux, gcc 4.2.1, 32 & 64 bit
  1. ICU4J
  1. x86 Linux, Sun Java 1.6, 32 bit
  1. Will send out an email to PMC with status of locale proposal.
  2. Incoming bugs:



2008/4/23 Agenda

  1. Locale keyword delimiter character (Steven, Yoshito)

  1. CLDR data integration activities in ICU (Steven)
  1. Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
  2. UCONFIG_NO_SERVICE=1 build issues on Cygwin/MSVC, http://bugs.icu-project.org/trac/ticket/6245 (Peter)
  1. OpenJDK Locale enhancement proposal (Yoshito)
  1. IUC32 - deadline April 25 (Friday) - http://unicodeconference.org/abstracts.htm  - require Bio and Abstract

  1. http://docs.google.com/Doc?docid=d3q3z35_4gkdmptc4&hl=en
  2. Steven: 2 abstracts in above doc - ICU Tutorial,  What's new in ICU
  3. Peter will write abstract on behalf of Mark for CLDR panel
  4. Addison and Mark will do BCP47
  5. Mark to ping Andy on Spoof detection
  6. Eric and Michael: posted to doc (collation, string search)
  7. Dave and Vladimir (to check on ICU on a diet)


2008/4/16 Agenda

  1. Please complete code review for 3.9.2!

  1. IUC 32 - April 25 deadline?
  1. Addison and Mark will do BCP47
  2. String search (Eric/Michael)
  1. Existing presentations? on Collation?  Laura's paper
  2. Mark- 'make it more general, unicode text sorting searching and matching, focus on string search'
  1. CLDR panel - often 2 sessions,  "presentation" + "panel discussion"
  2. send out abstracts this week or next- discuss offline any particular topics
  1. M3 update  - 3.9.3
  1. announce?  icu-support, icu-announce. tagged in svn
  2. summary - mark to edit here in the agenda
  1. ICU 3.9.3 supports unicode 5.1
  2. various fixes
  3. development, not final- although it has passed all tests.
  4. can point them to M1/M2/M3 links,
  5. can point them to buglist for [trunk of 3.8.1  vs. trunk of 3.9.3] - srl to do
  1. transliteration time bombs
  1. 4009 (thai-latin) - andy
  1. a page of code, puts a space inbetween each word before transliteration
  2. Andy to investigate.
  3. If it's not going to be in 4.0, withdraw and make it an RFE.
  1. 5608 (round trip in indics) - mark
  1. issue: when to support new chars in transliterations as r/t, not necessarily when they come out. when the chars are well supported in fonts (keyboards?) .  Is a CLDR issue (CLDR bug#1700? Mark to file for cldr1.6)
  1. need to be fixed, not high priority vis a vis CLDR
  1. Announcement text (draft) - Steven to add links
  1. We're please to announce the release of ICU 3.9.3, now providing support for Unicode 5.1. This is a development release, so is available through http://icu-project.org/repository/ [add info how to get]. For a full set of differences between this and the last official release (3.8.1), see YYY.
  1. Adding CLDR 1.6pre data into ICU?
  1. srl to get Rick to push the button to update data automatically
  2. mapping from users to status


2008/4/9 Agenda

  1. Another milestone tag for Unicode 5.1 integration? (Steven/Yoshito)

  1. Java coding style guideline update proposals (Yoshito)
  1. Unicode Conference (IUC 32) agenda
  1. April 25 deadline - http://www.unicodeconference.org/press-release-b.htm
  2. Ideas
  1. Latest version of ICU (Steven)
  2. Using ICU workshop (including packaging) (Steven, Peter-surrogate?)
  3. ICU on a diet (also talking about applications) (Vladimir, Dave)
  4. CLDR talk & panel (Steven, Mark, Deborah?/Chrish, ... put call out on CLDR list)
  5. Knotty problems in date/time parsing and formatting and timezones (Yoshito, Mark)
  6. Unicode spoof detection (Andy)
  7. String search (Eric/Michael)
  8. The right way to identify languages: BCP 47 (Addison, Mark) CONFIRMED
  1. Eric - collation


2008/4/2 Agenda

  1. Unicode Conference (IUC 32) agenda

  1. ICU4C project file - VS2008 (Andy)
  1. String search (Eric)


2008/3/26 Agenda

  1. Changes in tzdata2008b (Yoshito)

  1. logo: http://icu-project.org/~srl/iculogo.html   (Steven) (already released into the wild)
  1. Mark to dig up cat-eye version (Sauron's puppy)
  1. Unicode 5.1 status (Andy)
  1. Still working on it, a few things to get working
  2. Transliteration failures (Exhaustive tests that try every letter of the script - such failures are to be expected)
  3. C: Collation OK
  4. C/J: RBBI OK
  5. Property block failure (Mysterious)
  6. check sometime in the medium/near future.. decide about whether to roll M3 at that time
  7. Q: Could it be practical to update only Unicode data in the future? A: might be possible, probably not likely to accomplish this as just data.
     Optimizations stand against just changing via data.
  1. Perhaps segment: properties, collation, ..
  1. partial update may be possible in the future.
  1. data driven tests would at least reduce work between C/J
  1. ( According to Rick's email, estimated release for Unicode 5.1 is April 4 )

2008/3/19 Agenda

  1. 4.0M2 (3.9.2) status (Steven/Yoshito)

  1. Time slot: Markus to Switzerland soon, will revisit time slot in a few weeks.
  2. CLDR update
  1. Submitting opening soon
  2. will revisit schedule for data integration
  3. CLDR-driven code changes to ICU?  John to report back on what is or will be in CLDR that's not in ICU. File ticket on what is missing.
  1. plural
  2. duration
  3. interval
  4. calendar list support?
  5. spec changes: available formats, date formats..


2008/3/12 Agenda

  1. BCP 47 support (Mark)

  1. http://bugs.icu-project.org/trac/ticket/6204
  2. http://bugs.icu-project.org/trac/ticket/5767
  3. http://bugs.icu-project.org/trac/ticket/4229
  4. http://bugs.icu-project.org/trac/ticket/3730
  5. FYI: http://www.sil.org/iso639-3/codes.asp?order=639_3&letter=f
  6. http://www.iana.org/assignments/language-subtag-registry 
  7. Agreed to fix bugs that crop up to move towards BCP 47 support
  1. "digit" locale keyword (Yoshito)
  1. Java Locale th_TH_TH designate Thai digit to be used for numbers
  2. No locale mapping to ICU
  3. No definition in the current CLDR - http://unicode.org/cldr/bugs/locale-bugs?findid=1471 
  4. Past discussion - digit=[native|alternative] or digit=<script>
  1. wintz.c (Eric)
  1. windows registry key: http://support.microsoft.com/kb/914387
  1. Milestone 2 (Steven)


2008/3/5 AgendaNote:  Google people can not make the meeting today.  We can either postpone to next week, or schedule some other time to discuss specific issues.  -- Andy

This meeting was canceled.  -- Yoshito

2008/2/27 Agenda

  1. ISO-2022 and HZ non-shortest escape sequence (Peter)

  1. icu-project.org availability/upgrade (Michael)
  1. a


2008/2/20 Agenda

  1. ucol_getSortKey error handling (George)

  1. Changing the semantics of YEAR_WOY field? (Yoshito)

2008/2/13 Agenda

  1. Java SPI implementation packaging (Yoshito)

  1. DateFormatSymbols - getDateFormatBundle/initializeData (Yoshito/Doug)
  1. Coverity (George)
  2. Meeting time (Mark)
  1. Markus will be in Central European timezone for a while.
  2. Eric is in Hawaii.
  3. We may want to move the meeting earlier.
  1. Whitespace and syntax characters (Mark)

2008/2/6 Agenda

  1. ICU4C error behavior (e.g., UnicodeSet.isBogus) (Markus, George, Michael)

  1. Report from Unicode Symbols Subcommittee meetings during UTC (Markus)

2008/1/30 Agenda

  1. Confirm ICU 4.0 schedule after CLDR 1.6 discussion, set further intermediate dates (e.g., d01 & d02)

  1. Everyone agrees to postpone schedule by one month. (no disagreement)
  2. Milestone 2 postponed by 18 days.
  3. Milestone 2 reminder
  1. All tests should pass on reference platforms.
  2. Timebomb failures should be fixed.
  3. Good checkpoint to finish up reviews.

2008/1/23 Agenda

  1. ICU 4.0 schedule (Yoshito)

  1. Unicode 5.1 update status (Markus)
  1. Mutexes (Andy)

2008/1/16 Agenda

  1. ICU 4.0 schedule (Yoshito)

  1. Include CLDR 1.6 or not? CLDR vs. ICU schedules, not enough time.
  2. Tentatively move out ICU 4.0 release date to 2008-jul-02
  1. ICU features for new structure in CLDR 1.6 (Mark)
  1. Likely subtag structure from CLDR 1.5.1 supplemental data (Yoshito)

2008/1/9 Agenda

  1. Please review your tickets (George) http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=revw&order=priority&revw=%21

  2. What is the schedule for Vladimir's and Doug's tickets? (George)
  1. Is the ICU4J charset API able to go draft in 4.0 release? (George)
  1. Backwards string search (Eric)
  1. Discuss more offline
  2. Eric to schedule a meeting with Andy, Vladimir, Mark, Michael, Lloyd and Claire.
  1. CLDR bugs. (Mark)
  1. Report on meeting from yesterday
  2. If there are items in ICU that depend on CLDR features, we want to get those in by Tuesday.
  3. Mark to forward note that he sent.
  1. Copyright scanner (Eric)
  1. Noticed that the copyright scanner sometimes didn't work.
  2. Eric to run copyright scanner to see what else made it through the scanner.


Past Meetings


 

Process

 
Completed Action Items

  1. Number

  1. Due
  1. Done
  1. Who
  1. Description
  1. 52
  1. 20070418

  1. Yoshito
  1. Open a Trac ticket for ICU4J refactoring for better handling of deprecated warnings.  Carried Over from last week.
  1. 53
  1. 20070501

  1. George
  1. Add documentation of 20070425#3 (Data stability policy for ICU maintenance releases) for 20070502 meeting
  1. 54
  1. 20070701

  1. Mark
  1. Look at sizing for item 20070425#4, Option B
  1. 55
  1. 20070701

  1. Vladimir
  1. Look at sizing for item 20070425#4, Option C
  1. 56
  1. 20070501

  1. Andy
  1. Add note about string search bugs to user doc, send email to support/announce (in progress)
  1. 57
  1. 20070502

  1. Steven
  1. File bug to add full data version.
  1. 58
  1. 20070516

  1. George
  1. Work on icu-project.org site regarding ICU4JNI
  1. 59
  1. 20070516

  1. George
  1. Create related.html so that it mentions ICU wrappers and related projects.
  1. 60
  1. 20070516

  1. Everyone
  1. Send e-mail to icu-core about what M2 tasks are at risk of not making the milestone.
  1. 61
  1. 20070516

  1. George
  1. Setup link to ICU4PAS and send a reply
  1. 62
  1. 20070516

  1. Everyone
  1. Check task list for M2 release (duplicate of 60)
  1. 63
  1. 20070523

  1. George
  1. Readd ICU4JNI navigation links
  1. 64
  1. 20070530

  1. George
  1. Update PMC page
  1. 65
  1. 20070627

  1. George
  1. File bugs about perpetual timebombs
  1. 66
  1. 20070801

  1. Tex & George
  1. Work on PHP build instructions re threading, see 20070627 meeting minutes
  1. 67
  1. 20070801

  1. Yoshito
  1. CollectionUtilities to be moved from main ICU4J jar file to test package
  1. 68
  1. 20080123

  1. Andy H
  1. Find or recreate ICU4J charset tests





  1. dave to look at pragmatics of turning warning level down on windows header
  1. send announcement
  1. tz - update next week
  2. version # in src and data - already done
  3. update configure
  4. check for non ascii
  5. static code analysis
  6. code coverage - just generate report, dont' hold release for it.
  7. code review - have them in progress
  8. API docs - generate, post as interim 4.1 docs, post errors as informative
  9. tag - ask people not to do destructive changes a week before.
  1. already the policy to put destructive changes on private branches
  1. urename.h
  2. API change report -generate and send to core.
  3. send a note to the list that the release is tagged.
  4. run exhaustive tests
  1. NO
  1. status comments
  2. compare icu4j/jdk
  3. update license files
  4. verify memory allocation
  5. eclipse plugin
  6. testmap
  7. udraft.h
  8. XLIFF
  9. sample/demo
  10. data portability
  11. locale explorer
  12. readme
  13. userguide
  14. download page
  15. public site
  16. big announce release
  17. update trac
  18. post-release cleanup
  19. online information update
  1. check-in date - Sept 24th. Discuss in PMC 
  1. if there's an error: we will discuss, fix major things, …  will create a branch only if needed
  2. minimum: exhaustive tests pass on reference platforms
  3. fix what we can, document the rest
  1. Codepage Detection
  1. Google: needs better detection with short fragments of text such as queries.  currently needs about 100-200 bytes. Discuss offline. Perhaps keep the entire histogram, may need weighting.
  2. Eric is getting more queries about this.
  1. cross compilation
  1. icu 4.0 source (re)packaging - Steven to update BRS document.
  1. Security notice:
  1. ICU4C 4.0 resolves the vulnerabilities CVE-2007-4770 and CVE-2007-4771 which were found in earlier versions of ICU. The standard ICU tests verify that these have been corrected, however, the updated versions of the previous tests may be run by applying the following patch to ICU 4.0: r24324. As well, ICU4C and ICU4J 4.0 resolve the issue underlying CVE-2008-1036. [#6175, #6198]
  2. Steven to Update 4.0 download page to link to 6175/6198 and mention 2022, HZ

2008/9/3 Agenda

  1. Java 5 (Yoshito)
  1. sent notes to core -
  2. large subset of ICU4J is used by CDC
  3. Option 2 (subset) not viable
  4. Option 3: Keep 1.3 subset
  1. Preprocessor
  2. Important features for ICU:
  1. Generics
  2. Autoboxing
  3. varargs
  1. Unimportant:
  1. for loop
  2. enums
  1. Option 1: Fork
  1. Fork become maintenance mode
  1. Proposal: (option 4- default implementation Java 5) - 'continue current strategy'
  1. Add new APIs under //#ifdef
  1. does cause maintenance burden
  2. can remove the old-support once eclipse moves forward
  3. Mark volunteered to propose some Java 5 API change to ICU as test case
  1. Do not use some features (for:, enums would be difficult to map inside an ifdef)
  1. Yoshito will come up with proposal, 2wks
  1. (Q:JDK Locale integration?: A: continuing investigation)
  1. Meeting on Sept 11th instead of 10th?
  1. Thursday Sept 11th or Fri Sept 12
  2. editorial: 10:30-6 Thursday
  3. friday at google?
  1. 10:30-11:30 PT meeting, then lunch?
  2. count:   5 IBM, 
  3. Andy to send out directions
  1. bugs Incoming Bugs

2008/VIII/27 Agenda

  1. JDK 1.5 (Yoshito)
  2. J2ME still continuing, based on CDC FP 1.1, JSR219, JDK1.4
  1. class library issue known, but issue is whether more recent syntax/bytecode can be introduced
  2. specify bytecode 1.5, cannot be compatible with prior jvm
  3. Mark: option 1, code fork. >1.4 adds robustness to code and interfaces.
  4. already have conditional compilation
  5. not just functionality, but inheritance
  6. code fork means data needs to be maintained
  1. cldr <-> icu version association
  2. need to provide old data, or limit data that moves
  1. next eclipse - ICU 4.0, june 2009
  2. Yoshito: option 2, provide a small subset to meet CDC requirements, and then have a full ICU4J (4.2+, Java 5+, etc)
  1. what APIs/implementation are affected? Yoshito to determine. And the transitive closure of APIs that those apis call.
  2. subset would be a 1.3-compatible part of ICU trunk- no back port needed.
  3. then, determine if this is feasable.
  1. Mark on rotation in Zürich (Mark)
  1. move meeting to earlier? 10:00a PT?
  2. Effective Sept 17th. Steven to send out calendar invitations.
  3. No meeting Sept 10th, maybe 11th
  1. ICU Release Browser (Steven)
  1. http://apps.icu-project.org/icu4jweb/icuinfo.jsp
  1. http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uproj=icu4c
  2. http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uproj=icu4c&uvers=4.0 
  3. searchable by attribute: http://apps.icu-project.org/icu4jweb/icuinfo.jsp?uwhat=match&uattr=os%2Bversion&uval=AIX+5.3 
  1. http://icu-project.org/xml/icuinfo.xml 
  1. IBM specific testing: http://icu-project.org/xml/testing_ibm.xml
  1. phasing out 'reference' in public docs
  2. regularly vs. often doesn't mean anything
  3. regularly here means 'a binary is available'
  4. could add a 'distribution format'
  5. Q: 'if i pick up the source, did it pass?'
  6. Binary available, is us
  7. What is useful information
  1. binary available
  2. did tests pass on this platform ( yes- completely, unconfirmed [no reason to list this], known failure )
  1. link: 'if you are interested in helping to fix this platform'
  1. 'tested by: IBM'
  1. 4.2 reference platforms platform decisions?
  1. bugs Incoming Bugs
  2. Collation Sensitive Regex (Eric)

2008/8/20 Agenda

  1. Update ICU to require JDK 1.5, eg Generics in APIs.
  1. Eclipse team's note
  1. JFace (which uses ICU4J) currently targets CDC Foundation 1.0, and as far as I am concerned, should not be moved to 5.0 unless there is a compelling reason for doing so.  Se we would very likely keep using the particular subset of ICU4J we are using now, so that it should run with whatever version of ICU4J someone decides to bundle with their Eclipse-based product.  What we ship with the SDK is an important question of course.

    Also, to understand this a little bit better - as far as I know, there are parts of ICU4J that need to be updated regularly, as new currencies, time zones, and languages are introduced, correct?  If ICU4J decided to use Java 1.5 for upcoming releases, would it be possible to do these kinds of updates in a way that they benefit both the newest version of ICU4J and an old maintenance branch?  I guess I am asking if branching is really going to be that big of an overhead.
  2. This would push everyone except Equinox and SWT up to requiring Java 5. Since they are talking about after the 3.5 release, this would impact 3.6/4.0.  There was a lot of discussion about this in the context of e4 but nothing conclusive.
  3. No consensus, so it probably is worth bringing up in the Arch call.  I've included some of the responses I got to my query.

    Equinox doesn't seem to be too effected by the ICU plugin, but JFace is and probably wouldn't move up the the 5.0 EE.  Is it that people want to use 5.0 in the ICU API or that they want to use the 5.0 language features (like enum or foreach loops) within the compiled code?

    The Arch call is probably where they would have to think about what Eclipse itself will do.
  1. Eclipse platform's usage
  1. http://wiki.eclipse.org/ICU4J_Usage_in_Eclipse
  2. http://wiki.eclipse.org/index.php/ICU4J 
  1. Generic issues
  1. Sun Java SE 1.3/1.4 is out of support
  2. J2ME - http://java.sun.com/javame/reference/apis.jsp 
  1. Connected Device Configuration - Foundation Profile 1.1 (JSR 219) is based on JDK 1.4
  1. javac in JDK1.5 cannot produce Java bytecode works on JRE1.4 or older with source=1.5
  1. retroweaver can transform 1.5 bytecode to 1.4 - would not solve class library issue.
  1. Solutions
  1. (option 1) code fork
  1. code is effectively forked with 4.0 maint stream.
  2. issue: maintaining data going forward
  1. could only backport critical data changes.
  1. (option 2) separate "core" subset of ICU4J work on CDC1.0
  2. (option 3) keep Java 1.3 syntax
  1. revisit next week…
  1. ICU4J class file issue (2.6.1) - http://bugs.icu-project.org/trac/ticket/6505 
  1. Eclipse bugzilla - https://bugs.eclipse.org/bugs/show_bug.cgi?id=216147
  2. ICU meta file: http://icu-project.org/xml/icumeta.xml
  1. IUC32 status (Michael)
  1. Final slides due 8/25 now, send a note to Kevin if you did not receive any comments.
  2. Eric, Yoshito will be at IUC- no PMC at regular date/time Sept 10th
  1. move to thursday (Sept 11) or friday (Sept 12)?
  1. bugs Incoming Bugs

2008/8/13 Agenda

  1. bugs Incoming Bugs

2008/8/6 Agenda

  1. C trunk broken-?
  1. Need to swat - replace primitive types in C according to coding guidelines.
  1. yyyy, will add to CLDR, FlexibleDateTime stuff needs some fixes.
  2. C:  version updated to 4.1.1
  1. J already updated - data version to be updated - Yoshito to update today.
  2. C: translit r/t timebombs removed, bug filed in CLDR
  3. Agreed; remove old timebombs where possible. Yoshito to add to BRS
  1. review short bugs (monthly)
  2. Small design items in email
  1. Raising methods into Collator. Will have NOOP implementation; documentation will indicate that subclasses should support but don't need to; these are requests. Agreed that this should be done more generally; Mark to file concrete API changes on design list.
  2. PunyCode public. Mark to add Transform interface.
  1. Website:
  1. email: Proposed Demo page additions - Mark, Andy to do.
  1. Update ICU to require JDK 1.5, eg Generics in APIs. Eg use generics in APIs and implementations, allow us to have parallel APIs to JDK (right now we have holes). Options?
  1. Maintain 4.0 for Eclipse, split if necessary
  2. Have core that is unchanged (Mark's comment - doesn't meet requirements)
  1. Bugs

2008/7/30 Agenda

  1. Schedule for Ticket #6426, excessive memory use in TimeZone Formatter (Peter)
  1. Apple needs by 9/30 (and needs to roll back to our 4.0 sources), can do some or all of the work. Discuss possible approaches?
  2. Actually, Markus already checked some fixes into trunk that reduce the size of the parsing data structure and go most of the way to addressing this.: That is Ticket #6414. That is probably enough to address the Apple issue, so that #6426 can be for more long-term solutions such as lazy initialization.
  3. Andy - another related issue, resources not flushed from the resource bundle cache in ICU4C. He is working on some code to address this. He will create a separate ticket for this, it is a general issue.
  4. http://bugs.icu-project.org/trac/ticket/6414 should improve the memory footprint.
  1. Performance tests/metrics for new features (Peter)
  1. Some way to instrument tests to measure peak memory usage during test? Needs investigation
  2. More testing of memory usage of development releases of ICU
  3. Peter will create a ticket
  1. https://bugs.webkit.org/show_bug.cgi?id=15914
  1. Implement Unicode functionality using GLib.
    This could be a fairly easy way to shave off 10M shared library memory use due to ICU...
  1. TimeZone#setDefault problem
  1. Yoshito sent the problem description to the core ML.
  2. If no objections, Yoshito will make the change as described in the message.
  1. Bugs Incoming Bugs

2008/7/23 Agenda

  1. trivial bugs (continued from last time)
  1. from last week - some bugs are trivial, take only a little bit of time to do
  2. we'll have spent more time looking at the bug than fixing it
  3. "fix-it day", to just fix bugs
  1. reminder message
  2. bugs by owner, 4.1.1 
  3. Bugs by owner, 4.2
  1. need a way to have these bugs show up as higher priority, especially earlier in the release.
  2. These are (trivial || minor && low weeks) bugs.
  3. Tactical: Mark all bugs which are quick to do, and then review them.
  4. Review once a month, at the start of the month
  1. ICU functional area contacts
  1. ICU Functional Area Contacts 
  1. Go over these, make sure appropriate
  1. Aligned to the current trac component keywords - are the current keywords appropriate?
  2. ICU 4.0 components assignment - for comparison
  3. split up formatting?
  1. numbers (currency, ...)
  2. dates (+timezone)
  3. "other"?
  1. Ticket Priorities and Dispositions
  1. Priorities:
    blocker — Item prevents shipment of ICU and/or other major tasks to proceed.
    critical — Major crashing bug or a feature whose absence prevents usability.
    major — Important feature or bug fix.
    +medium — Bug of medium importance.
    minor — An item which would be good to do, but isn't very important. Do if there is time.
    trivial — An item of very little importance and impact, such as a minor typo in documentation.
    zero — An item which is valid, but not currently considered important enough to fix or do. These items may be left open as zero instead of closed as 'wontfix' so they are searchable in case their priority becomes important later.
    assess — The assignee of this item is requested to determine what the bug priority is, if any. It may be that the item is not valid or should not be done.
  2. Dispositions:
    Tickets may be closed with the following dispositions:
    fixed — Only the person listed as "reviewer" for this bug should close the bug as fixed, once they have reviewed the code.
    invalid — The ticket is not valid, it is not a bug or feature request. It may be spam, or just useless text.
    wontfix — This will not be fixed (or implemented) by ICU.
    duplicate — The bug has already been filed.  Add a comment to the duplicate bug saying, "This bug is a duplicate of #1234".     If the duplicate adds anything to the discussion, add a comment to the remaining bug such as, "Bug #4321 was closed as a duplicate, but it contains some interesting test cases."
    worksforme — The bug was not able to be reproduced.
  3. Issues
  1. Gap between major and minor - lots of 'medium' category.
  2. worksforme conflates  "this was broken before but now fixed",  "could not reproduce this, even in an earlier version"
  1. "moot" - "does not matter anymore" (could refer to functionality that is no longer supported)
  2. Must close worksforme with a comment, "works in C4.8" or "works fine in r43218"
  3. could add a field: "works in r12345" or "fixed in 4.8"
  4. Could add a button, "which releases have this code"
  1. action items- DONE.
  1. remove blocker
  2. add medium
  3. drop trivial
  1. 4.2 planning
  1. Review each company's load and move some tickets across companies.
  1. example: may have Google:100, but under Yoshito. If IBM has no interest, transfer to Google person.
  2. Query: Bugs with a Load set and no reviewer
  1. bugs Incoming Bugs

2008/7/16 Agenda

  1. tzdata2008d
  1. Uploaded the files to the ICU project server (zoneinfo and metazoneInfo)
  2. Updated ICU4J Time Zone Update Utility to support metazoneInfo.
  3. supplementalData which contains zone aliases and region mapping data is not included, because the single version does not work well with all ICU versions.  We agreed this in the last PMC meeting.  supplementalData was updated along with zoneinfo/metazoneInfo in ICU4C/J SVN trunk only.
  4. Apple has a build system problems with the CLDR dependency - metazoneInfo (and supplementalData), see next item
  1. Need to be able to update Olson data - and now metazoneInfo - independently of ICU (Apple)
  1. In Mac OS X, the Olson data is used directly by many projects: ICU, BSD, etc.
  2. All of those projects need to be in sync, using the same version of Olson data. We need to be able to update the Olson source data and and all dependent source data (which now includes the metazone data) independently of ICU sources. We have a TimeZoneData project that contains Olson data and is submitted independently of ICU.
  3. We cannot use zoneinfo.res and metazoneInfo.res. The Apple build process requires that we submit sources, and moreover, the other clients of the time zone data do not understand zoneinfo.res.
  4. The ICU build process does permit using out-of-source-tree Olson data to override the data in the ICU source tree. If the makefile variable TZDATA points at tzdatayyyyv.tar.gz (a version of the Olson data), source/data/Makefile.in will override the zoneinfo.txt in source/data/misc. It does this by running the tools in source/tools/tzcode to take the Olson data pointed to by TZDATA, and producing a new version of zoneinfo.txt that gets used instead of the one checked into the source repository. In our case, TZDATA points at /usr/local/share/tz/tzdatayyyyv.tar.gz, which is created by the TimeZoneData project. So for example, in Mac OS X 10.5.4 we have the following in /usr/local/share/tz/:
  1. -rw-r--r--   1 root  wheel  190412 Oct  1  2007 tzcode2007h.tar.gz
  2. -rw-r--r--   1 root  wheel  162187 Oct  1  2007 tzdata2007h.tar.gz
  1. Although 10.5.4 is using ICU sources derived from ICU 3.6 (which had Olson 2006j), and those have been modified for Apple's Windows builds to have zoneinfo.txt with Olson 2007k, when built for Mac OS X, ICU will pull the Olson 2007h data from /usr/local/share/tz.
  2. What we need now is add metazone source data (txt or xml) to this process. We need to be able to put the relevant source files in the TImeZoneData project, and the ICU build needs to allow metazone source data in some place like /usr/local/share/tz/ to override the metazone info in the ICU sources.
  3. => Agreed in principle to makefile changes that allow metazoneInfo.txt* from outside the ICU source tree to override the one in the ICU source tree, Peter will file a bug and submit proposal to icu-design. (* or whatever metazone-related files fall out as a result of http://bugs.icu-project.org/trac/ticket/6444 ).
  1. Trac system usage issues
  1. Use of "accept" - ignore "accept" - used for nothing
  2. Priority for enhancements vs. implementation bugs
  3. How to handle trivial bugs
  4. Usage of Load - 100% means intention to complete the bug; doesn't mean fall-on-your-sword commitment. May have multiple orgs with 100%. 50% means willing to work on it and/or trade off working on that with other 50%.
  5. Add "reporter" to spreadsheet - Andy to do.
  6. Automatic ticket assignment - a new incoming bug will be reassigned to functional area expert's bucket automatically
  1. Yoshito's comment - I like to go through incoming bugs in weekly PMC meeting and assign each of them to developers at that time.  A new incoming ticket may be high priority for some companies, but could be low priority for other companies.  If a ticket is assigned to a developer who belongs to the latter group, people in the former group may miss it.  In major release planning phase, each company goes through the whole ticket list once (like we're doing now).  But once this is done, we rarely inspect all of open tickets.  I have no problem to create the functional area assignee list - when none of PMC member companies are interested in a new incoming ticket, the ticket may be moved to the functional area assignee.
  2. Shared doc - developer per each functional area -> Yoshito will create the google spreadsheet.
  1. String search?
  2. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/7/9 Agenda

  1. 4.2 schedule
  1. more realistic dates - Jan 1 is not a likely ship date
  2. decree a milestone when a significant feature is done
  3. milestone prevents us from drifting, and reduces catch-up at end (when too many items are deferred to the GA timeframe)
  1. do some BRS items  (LRS)
  1. use branches - keep the trunk more stable.
  2. treat the milestones as real deadlines - some lead time before release date, ~1 wk
  1. could use maint-4-1? or just don't put small fixes in.
  2. code slush to maint-4-1 a week before each milestone release, and then BRS
  1. 4.2 planning
  1. Policy for critical/major items
  1. unscheduled critical/major are being moved to assess..
  2. please either move to minor/trivial (unscheduled), or set a milestone 4.2 or 4.4
  3. Names: "trivial" sounds like an amount of work.  Is it a Severity?  Priority: ROI. Severity: Urgency+Fixability. (applies to bugs, not features.)
  4. trac mainpage links to: http://www.icu-project.org/processes/ticket-lifecycle.html 
  5. We may need more explanation for each keyword. The above page is a good place to document this.
  6. Steven and Yoshito to draft update with explanation for keywords
  1. 7/23 - 4.2 candidates review
  1. fill in Load:   IBM:50 Google:100 Apple:100 etc ( % of interest&&availability to work on the bug)
  1. tzdata 2008d status (Yoshito)
  1. metazone mappings change (otherwise, fallback mapping will occur)
  1. (olson data) -> zoneinfo.txt/.res
  2. (CLDR metazoneInfo.xml) -> metazoneInfo.txt/.res
  3. (CLDR supplementalData.xml) -> supplementalData.txt/.res (contains other unrelated things..)
  4. proposal1: include zoneInfo and metazoneInfo but not supplemental data.
  5. proposal2: separate data out so that timezone-only .res files are created. (so that only those are updated.)
  1. CLDR 1.6.1
  1. checked in
  1. Update instruction change
  1. zoneinfo.res, metazoneInfo.res, (skip supplementalData.res?)?
  1. ICU4J Time Zone Update Utility changes
  1. needs to be updated to install metazoneInfo.res
  1. hope to wrap up by end of week or early next
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/7/2 Agenda

  1. 4.0 - ready to ship? (steven)
  1. translit issue - give it 2 hours to investigate
  2. Draft 4.0 release page:  http://icu-project.org/download/4.0.html
  1. 4.1 - update version in trunk? (yoshito)
  1. may be timebomb failures
  1. 4.2 schedule
  1. From 6/11 meeting: "May 1 GA" - OK.
  2. Need intervening milestones
  3. CLDR: OK with CLDR release in the May timeframe.
  4. No Unicode update in timeframe. - 5.2 in fall?
  5. n milestones, 3mo apart - adjust based on interesting features
  1. M1 Oct 1, 2008
  2. M2 Jan 1, 2009
  3. M3 March 15 - DCUT API Freeze
  4. M4 Apr 1, 2009 
  5. M5 April 15 - code freeze
  1. GA May 1 2009
  1. Please clean up 4.0 tickets
  1. 40+ reviews
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
  1. in 2-3 weeks - go over 4.2 bugs, determine who is interested..
  1. 6397 yoshito
    6398 andy
    6399 xji
    6400 yoshito
    6408 doug
    6409 yoshito
    6364  srl
    6383 srl
    6389 doug
    6390 doug
    6394 andy
    6410 yoshito
  2. July 23rd - go over 4.2 bugs, determine who is interested in what, set 'load' field.
  1. "interest" here means "you are willing to work on it".
  1. Transliterator bug…  user error.
  1. \u3051\u30c0  -> Keダ  ( should be 'keda'
  1. works from a custom intltest test
  1. \u03cc  NFD should change this char
  2. Due to a bad custom rule named 'NFD'. Filed bug 6417..

Old 4.0 situation room ( moved on July 9th, 2008  by srl )

  1. john- cldr 1.6d03 (RC)
  1. #6378
  1. Integrate Data from CLDR release-1-6-d03
  1. data
  1. Current
  1. 4.0
  1. defect
  1. blocker
  1. 06/23/08
  1.     -
  1. C: r24307 ticket:6378: Integrate CLDR release-1-6-d03
  2. J: r24308
  1. xiaomei - interval (c /j DONE?)
  1. #6157
  1. C: r24280
  1. doug- plural
  1. J: #6391 r24290
  1. claire - plural
  1. C: #6374:  r24299, r24310
  1. dave- resource data
  1. comes automatically?  Yes.
  2. J: #6124 r24287
  3. C  #6149  r24285 typos in comments??
  1. srl- cygwin mingw build changes #6322 (possible), possible icu4j doc
  1. C #6098 ( version, release, etc )
  1. C r24291 update readme, add API change report.
  1. mark- crashing bug in translit. Mark wants to fix: 
  1. #4200
  1. Transliterator crashes on certain data
  1. transliterate

  1. 4.0
  1. defect
  1. critical
  1. 11/03/0

ICU 4.0 Situation Room

 release schedule

2008/6/25 Agenda

  1. Tex @ NetApp - http://icu-project.org/projectinfo.html 
  1. (I see you are on the document, good. -srl)
  1. 4.0 status (Steven/Yoshito)
  1. CLDR status (John)
  1. 1.6d02 - into icu 4.0d03
  2. some further changes..
  1. az
  2. weekend
  3. korean
  1. close..
  2. probably final @ CLDR meeting 7/1/08
  3. plural rule format change
  1. plurals don't have getavailablelocales ( or getfunctionalequivalent ) - doug to file api
  2. isolated change - plurals.txt
  1. 1.6d03? (RC?) - probably tomorrow 6/26
  1. Any tickets not yet merged into 4.0 release stream?please update w/ bug number - above list, situation room
  2. trunk is post-4.0 effectively,  notify icu-core (for srl and yoshito) with revision/ticket for merge into branches/maint/maint-4.0 which will be re-tagged -
  3. Code review - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=revw&order=priority&revw=%21
  4. All please check d03,  download page, readmes.
  5. icu4c 4.0d04 (RC) - EOB 6/27? 
  1. see 'situation room' above
  1. API question: PluralRules.select vs. PluralFormat.format (Yoshito)
  1. can change after 4.0 - @draft 3.8
  2. (long)1 vs (double)1.0
  3. future: need to look at  "1 hour" vs "1.0 hours"
  4. format (# of decimals) decoupled from primitive type (long vs. double)
  5. only on java
  6. bug #?
  7. Doug will take bug #
  8. format method is final
  9. send exact API changes to icu-design (NB: api docs will need to be regenerated)
  1. bugs http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/6/18 Agenda

  1. Welcome Brian Rower (IBM)
  2. 4.0 status (Steven/Yoshito)
  1. Any remaining items to be included in 4.0?
  1. Please start cleaning up your 4.0 tickets - move them to 4.2/UNSCH if you have no plan to work on them for 4.0
  2. Any non-trivial tickets still working?  Please let Steven/Yoshito know about them.
  1. (you can list them here...)
  2. EBCDIC issues in progress (mostly test issues)
  1. late API changes are on icu-design - interval format parameter, UCharacter
  1. CLDR status update
  1. Dave has merged changes into trunk
  2. two upcoming merges:  one targetting the branch (today/tomorrow), and one later one.
  1. Reminder - creating branch the end of tomorrow (Thu June 19)
  1. will proceed with ICU d03 on June 23 as planned, but know there may be some cldr-related churn
  2. (CLDR: tag d01 at end of day after default content resolved?)
  1. Please work on code reviews!
  2. BRS items - any problems so far?
  1. environment tests (eric)
  1. timezones on mac - gmt offset not a multiple of :15 - saudi arabia, africa, ..
     - eric's note subj: '[icu-core] environmentTest Timezone Failures' 6/12/08 9:27 PM PT.  Could make this a warning in this or future release. Historical zones.
  2. ko_KR.CP949 locale / backslash / ucbuf  ..  default codepage handles "\" or "\u" incorrectly.  Fixed in most places, but difficult with included resource. "\uXXXX"  -> something like "₩uXXXX".
  1. recommendation: map 0x5c <-> U+005c
  2. action: verify that it is not a regression from 3.8.1. Document in readme. Investigate in 4.2.
  3. main() { puts(ucnv_getDefaultName()); } will return default codepage - see what is returned under ko_KR.CP949
  4. post 4.0:  prefer the default converter that DOES map 0x5C <-> U+005C
  1. test without data - done (claire)
  2. no namespace  - needs to be added to BRS (andy)
  3. uconfig test (srl) - usearch will require brkiter
  1. 4.0 download page
  1. Steven to update readme with a link to generate a list of bugs fixed
  1. CLDR default content wrap up

http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/6/11 Agenda

  1. 4.0 Schedule (Steven)
  1. Branch - Scheduled for June 13th
  2. but CLDR not ready?
  3. Dave made changes to CLD
  4. Changes to LDML2ICUConverter tool for data differences
  5. Dave will be checking updated data into a branch, pull into icu4j
  1. plural rule on icu side waiting to be hooked up
  1. CLDR merge: Mon, Jun 16th - then some bug fixing
  2. Branch: end of Thu June 19th, and communicate on Fri, Jun 20th any issues -  Steven and Yoshito will merge as needed
  3. d03: Mon Jun 23
  1. 4.0 BRS
  1. http://www.icu-project.org/processes/release.html 
  2. http://spreadsheets.google.com/ccc?key=plRXpLeUpVd2kRTY-HJGEFg
  3. look at bugs with code and no reviewer http://bugs.icu-project.org/trac/report/18
  1. API Tweaks - Xiaomei interval format
  1. small, will send a list to design list
  1. 4.2 Schedule
  1. Steven's concern: March ship brings end of release at end of year
  2. Issue: CLDR schedule?  Gap with ICU vs CLDR brings unexpected difficulties
  3. CLDR schedule not decided
  1. longer candidate timeframe
  1. If the ICU M release is usable, will be some value
  2. at least 1 or 2 rigorous milestones - 
  1. end of March/beginning of April?
  2. more development on branches
  1. develop milestones in branch
  2. could have maint-4.1 
  1. keep the trunk clean - continuous integration?
  2. may want more policy: keep things in branches unless it is in shape, with test coverage
  3. Andy's Mark's Suggestion, for large things, review on branch before merge?
  1. IBM would like to see release in June?
  2. Tentative GA: May 1
  1. Withdraw spoof detection. #4790 (Andy)
  1. in a branch, headers checked in
  2. will withdraw headers for 4.0
  3. verify that net result is no change in trunk, and change bug to 4.2

(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/6/04 Agenda

  1. Bug 6321, incorrect short time zone names.  Hot, I'm looking at it but, could use help.  Snarled in metazone code.  John?  (Andy)
  2. Encoding Selection (Andy)
  1. makes heavy use of STL - not in the form to go in now.. involves memory management
  2. pushed out to 4.2
  1. 4.0 Schedule (Steven) 
  1. http://www.icu-project.org/processes/release.html updated
  2. spreadsheet: status
  3. Will send out task items, to be assigned today and tomorrow
  1. CLDR 1.6 Schedule (Peter)
  1. CLDR 1.6 BRS will start June 10th
  2. John is already running tests
  3. Expect data to change until end of next week
  1. 4.2 tentative schedule (Steven)
  1. Slush: Feb 4, 2009
    DCUT: Mar 4, 2009
    GM: Apr 29, 2009
    GA: May 13, 2009
  1. CLDR Plan: open up Survey Tool right after 1.6
  1. need data firming in Indic
  1. Won't be a major Unicode release until later
  2. Mark: end of March? to avoid holidays. End of quarter is good

2008/5/28 Agenda

  1. Code reviewers for String Search: Mark & Andy? (Eric)
  2. We have an internal customer for ticket:2469 (Regexp, support for Unicode Regexp Level 2) and would like to see it scheduled for 4.2. (Eric)
  1. VS 32/64 bit -
  1. if the intermediate and output dirs are different, can have simultaneous build.  Eric to file bug based on support mail .
  1. 4.0d02 status (Steven, Yoshito)
  1. seems to be building clean on C after fixes
  2. J: errors?
  3. any pending changes?
  4. tag 3pm PT today?
  5. TimeUnitFormat - in J (6303), but won't make it in C for 4.0. Please move/refile bugs for 4.2
  1. 4.0 Release Items: Yoshito will post worksheet after this meeting.
  2. 4.0 reviews
  3. Copyright - Steven is checking on status

(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

 

2008/5/21 Agenda

  1. 4.0d01 status (Steven, Yoshito)
  1. can check in code at the same time as design proposal
  2. ICU4J binary/source compatibility:
  1. nothing stated
  2. many people run against the prebuilt jars
  3. userguide should state the policy
  1. future: discuss clear statement on icu4j stability runtime vs. source.
  2. cutoff for DCUT tag- today, 3pm pacific - non functional stubs OK
  1. pending for #6291 - peter
  2. charset selection (design posted today)
  3. date intervals
  4. simple durations
  1. delay dcut 1 week  for charset selection/date intervals/simple durations?
  1. could still go out with d01 but not call it the final DCUT
  2. still get things in by 3pm if possible (especially changes to existing APIs)
  1. CLDR data generation status
  1. John and Dave to coordinate getting consistent results from LDML2ICUConverter
  1. Welcome Jason Spieth
  2. Transliterator timebomb status (4.0)
  1. Thai-Latin Instantiation seems to be fixed for C  (  ( no data test yet ) 
  1. Not the case for J... ?  - was already in J
  1. unicode r/t tests still were failing, moved to post 4.0 timebomb
  1. If this is not fixed by CLDR 1.6/ICU 4.0, then these tests should be removed, and moved into CLDR (post 4.0)
  1. date format pattern generator timebomb, J, C?
  1. Yoshito will send a note out
  1. API stability status update
  1. ICU4J (Yoshito)
  1. API Change report for ICU4J 4.0 -> http://icu-project.org/~yoshito/api40/icu4j_compare_381_400.html
  2. Charset - @draft ICU 3.6/3.8 are promoted to @stable except CharsetCallback
  3. BiDi / Plural format / Relative date format / Duration format / Time zone transiton / Rule based time zone - keep @draft ICU 3.8
  4. GlobalizationPreferences - keep @draft ICU 3.6
  5. UnicodeSet.XSymbolTable - keep @draft ICU 3.8
  6. Indian/Taiwan calendar - changed from @draft ICU 3.8 to @stable
  1. ICU4C (Steven)
  1. analysis not done
  2. will check in and announce changes for discussion today
  1. ICU 4.0 release tasks (Yoshito)
  1. will consolidate tasks, and make copy for each version
  2. discuss next PMC
  1. ICU4J Locale Service Provider for Java 6 (Yoshito)
  1. Separated root - icu4j/localespi
  2. Separated jar file - icu4j-localespi.jar

(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

2008/5/14 Agenda

  1. IUC32 status?
  2. New locales for 4.0 ( e-mail from John )
    ii_CN - Yi (China)
    fr_SN - French (Senegal)
    ha_Latn_NG - Hausa (Latin, Nigeria)
    pa_Arab_PK - Punjabi (Arabic, Pakistan)
    zh_Hans_HK - Chinese ( Simplified, Hong Kong )
    zh_Hans_MO - Chinese ( Simplified, Macao )
    zu_ZA - Zulu ( South Africa )
    ne_NP - Nepali ( Nepal )
    ne_IN - Nepali ( India )

    No objections noted.  If anyone has any additions/deletions, please contact John.
  3. ICU 4.0 status
  1. API freeze coming up next week - May 21
  2. Will update version to 4.0 at DCUT (may 21 ,D01) as well as data version to '40'
  1. this affects timebombs- 5608, 4009 -
  1. CLDR Integration status
  1. John has snapshot with all tests working on the C side.  Will integrate to trunk within next day or 2, followed by building/integrating data for J.

(as time permits) Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody

 

2008/5/7 Agenda

  1. Locale/Language Fallback and Java Locale proposal (Yoshito, Mark)
  1. http://docs.google.com/Doc?docid=dfqr8rd5_66cmjvtgfn&hl=en (Mark's Locale/Language Fallback)
  2. http://docs.google.com/View?docid=d3q3z35_10g9wwphg6 (Yoshito's JDK locale enhancement proposal)
  3. bcp47
  1.   The relevant tickets are:
       http://bugs.icu-project.org/trac/ticket/6204   (4.0)
       http://bugs.icu-project.org/trac/ticket/6177 (4.0)
       http://bugs.icu-project.org/trac/ticket/5767 4646 support ( not scheduled )
  2. cldr bcp47 syntax: http://unicode.org/cldr/bugs/locale-bugs/docs?id=1730 
  1. LDML2ICUConverter CVS tag (Mark)
  2. ...
  3. Design Proposals - Dave - ticket 6228

2008/4/30 Agenda

  1. CLDR Data integration status (John)
  1. tool issues (LDML2ICUConverter, LocaleMap null) (Mark)
  2. change in supplemental / plural?
  1. plurality of supplemental parsers (Peter to look at, may be a few days before he can)
  1. John back in town tomorrow 5/1 - John to send out status of integration
  2. Dave investigating which codebase was 1-5-1, not tagged (1.5.1-RC was tagged)
  1. IUC 32 abstract submission status
  1. submitted…
  1. JDK Locale proposal (Yoshito)
  1. http://docs.google.com/View?docID=d3q3z35_7c2rcqwg4 (older version as of this writing, will be updated)
  2. Had some meetings to incorporate feedback
  3. plan is to send to openjdk i18n list sometime next week, more meeting(s) next week
  4. to give overview later in this meeting?
  1. 4.0 status (Yoshito, Steven)
  1. http://bugs.icu-project.org/trac/milestone/4.0
  2. 3 weeks until API freeze / D01 May 21st - at least API proposals and stub code. Keep this schedule.
  1. spoof detecton on branch (Andy?)
  2. likelysubtags on branch - waiting on LDML - API4C submitted to design, API4J to be submitted (Dave)
  3. Interval/Duration/Plural has API (Mark?)
  1. Please keep an eye on reviews and bug assignments.
  1. 4.0 Reference Platform (Steven, Yoshito)
  1. IBM's reference platform list
  1. Java
  1. AIX 6.1, IBM Java 5 (64-bit)
    Solaris 10, Sun Java 5 (32-bit)
    Windows Vista SP1, IBM Java 5 (32-bit)
  1. C/C++
  1. AIX 6.1, 64 bit VA 9
    Solaris 10 Sparc 64?, Sun Studio 12
    Red Hat Enterprise Linux 5, x86-32? GCC 4.1.2 (Feb 13, 2007)
    HP 11iv3 - aCC A.6.14
    Windows Vista SP1 32 bit, Visual Studio 2005
    Windows Server 2003 64 bit, Visual Studio 2005
  1. Apple builds & tests for the following:
  1. Darwin / Mac OS X 10.5, PowerPC 32/64 & x86 32/64, gcc 4.0.1
  2. Windows XP SP2, x86/32, Visual Studio 2005 (compiler & linker) + Cygwin (make)
  1. Q: Reference list  including platforms from other parties?
  1. IBM's list driven by IBM's reference support.  IBM-ism.
  2. reference platform list (for C) [historically] determines what platforms binaries are built for.
  1. We do build on other platforms
  2. http://source.icu-project.org/repos/icu/icu4j/tags/release-3-8-1/readme.html#PlatformDependencies
  3. http://source.icu-project.org/repos/icu/icu/tags/release-3-8-1/readme.html#HowToBuildSupported
  1. Public only cares about (a) which platforms are tested, and (b) which platforms are available for binary download.
  1. could have a page for people to report in which platforms build OK
  2. could semi-automate this with configure generating a 'report' to send in
  3. General Public does not care about reference vs. tested
  1. Proposal:
  1. public site: list of tested versions,  list of downloadable versions
  1. change 'reference' to 'regularly tested'
  2. list 'binaries are available' for the list where binaries are available
  1. Vista 32 binaries probably good for w2k, xp, 2003…
  1. internally: keep track of who tests which platforms
  1. which PMC members build which platforms regularly
  2. which PMC members build which platforms occasionally
  3. which external people claim platform X works on ICU  #.#
  4. which external people have claimed in the past ICU worked on some platform ('rarely tested')
  5. … combine these into one external list for the ICU web page
  1. Google's list of regularly tested platforms

ICU4C

  1. x86 Linux, gcc 4.2.1, 32 & 64 bit

ICU4J

  1. x86 Linux, Sun Java 1.6, 32 bit
  1. Will send out an email to PMC with status of locale proposal.
  2. Incoming bugs:

2008/4/23 Agenda

  1. Locale keyword delimiter character (Steven, Yoshito)
  1. CLDR data integration activities in ICU (Steven)
  1. Incoming tickets - http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=status&order=priority&owner=somebody
  2. UCONFIG_NO_SERVICE=1 build issues on Cygwin/MSVC, http://bugs.icu-project.org/trac/ticket/6245 (Peter)
  1. OpenJDK Locale enhancement proposal (Yoshito)
  1. IUC32 - deadline April 25 (Friday) - http://unicodeconference.org/abstracts.htm  - require Bio and Abstract

  1. http://docs.google.com/Doc?docid=d3q3z35_4gkdmptc4&hl=en
  2. Steven: 2 abstracts in above doc - ICU Tutorial,  What's new in ICU
  3. Peter will write abstract on behalf of Mark for CLDR panel
  4. Addison and Mark will do BCP47
  5. Mark to ping Andy on Spoof detection
  6. Eric and Michael: posted to doc (collation, string search)
  7. Dave and Vladimir (to check on ICU on a diet)

2008/4/16 Agenda

  1. Please complete code review for 3.9.2!
  1. IUC 32 - April 25 deadline?
  1. Addison and Mark will do BCP47
  2. String search (Eric/Michael)
  1. Existing presentations? on Collation?  Laura's paper
  2. Mark- 'make it more general, unicode text sorting searching and matching, focus on string search'
  1. CLDR panel - often 2 sessions,  "presentation" + "panel discussion"
  2. send out abstracts this week or next- discuss offline any particular topics
  1. M3 update  - 3.9.3
  1. announce?  icu-support, icu-announce. tagged in svn
  2. summary - mark to edit here in the agenda
  1. ICU 3.9.3 supports unicode 5.1
  2. various fixes
  3. development, not final- although it has passed all tests.
  4. can point them to M1/M2/M3 links,
  5. can point them to buglist for [trunk of 3.8.1  vs. trunk of 3.9.3] - srl to do
  1. transliteration time bombs
  1. 4009 (thai-latin) - andy
  1. a page of code, puts a space inbetween each word before transliteration
  2. Andy to investigate.
  3. If it's not going to be in 4.0, withdraw and make it an RFE.
  1. 5608 (round trip in indics) - mark
  1. issue: when to support new chars in transliterations as r/t, not necessarily when they come out. when the chars are well supported in fonts (keyboards?) .  Is a CLDR issue (CLDR bug#1700? Mark to file for cldr1.6)
  1. need to be fixed, not high priority vis a vis CLDR
  1. Announcement text (draft) - Steven to add links
  1. We're please to announce the release of ICU 3.9.3, now providing support for Unicode 5.1. This is a development release, so is available through http://icu-project.org/repository/ [add info how to get]. For a full set of differences between this and the last official release (3.8.1), see YYY.
  1. Adding CLDR 1.6pre data into ICU?
  1. srl to get Rick to push the button to update data automatically
  2. mapping from users to status

2008/4/9 Agenda

  1. Another milestone tag for Unicode 5.1 integration? (Steven/Yoshito)
  1. Java coding style guideline update proposals (Yoshito)
  1. Unicode Conference (IUC 32) agenda
  1. April 25 deadline - http://www.unicodeconference.org/press-release-b.htm
  2. Ideas
  1. Latest version of ICU (Steven)
  2. Using ICU workshop (including packaging) (Steven, Peter-surrogate?)
  3. ICU on a diet (also talking about applications) (Vladimir, Dave)
  4. CLDR talk & panel (Steven, Mark, Deborah?/Chrish, ... put call out on CLDR list)
  5. Knotty problems in date/time parsing and formatting and timezones (Yoshito, Mark)
  6. Unicode spoof detection (Andy)
  7. String search (Eric/Michael)
  8. The right way to identify languages: BCP 47 (Addison, Mark) CONFIRMED
  1. Eric - collation

2008/4/2 Agenda

  1. Unicode Conference (IUC 32) agenda
  1. ICU4C project file - VS2008 (Andy)
  1. String search (Eric)

2008/3/26 Agenda

  1. Changes in tzdata2008b (Yoshito)
  1. logo: http://icu-project.org/~srl/iculogo.html   (Steven) (already released into the wild)
  1. Mark to dig up cat-eye version (Sauron's puppy)
  1. Unicode 5.1 status (Andy)
  1. Still working on it, a few things to get working
  2. Transliteration failures (Exhaustive tests that try every letter of the script - such failures are to be expected)
  3. C: Collation OK
  4. C/J: RBBI OK
  5. Property block failure (Mysterious)
  6. check sometime in the medium/near future.. decide about whether to roll M3 at that time
  7. Q: Could it be practical to update only Unicode data in the future? A: might be possible, probably not likely to accomplish this as just data.
     Optimizations stand against just changing via data.
  1. Perhaps segment: properties, collation, ..
  1. partial update may be possible in the future.
  1. data driven tests would at least reduce work between C/J
  1. ( According to Rick's email, estimated release for Unicode 5.1 is April 4 )

2008/3/19 Agenda

  1. 4.0M2 (3.9.2) status (Steven/Yoshito)
  1. Time slot: Markus to Switzerland soon, will revisit time slot in a few weeks.
  2. CLDR update
  1. Submitting opening soon
  2. will revisit schedule for data integration
  3. CLDR-driven code changes to ICU?  John to report back on what is or will be in CLDR that's not in ICU. File ticket on what is missing.
  1. plural
  2. duration
  3. interval
  4. calendar list support?
  5. spec changes: available formats, date formats..

2008/3/12 Agenda

  1. BCP 47 support (Mark)
  1. http://bugs.icu-project.org/trac/ticket/6204
  2. http://bugs.icu-project.org/trac/ticket/5767
  3. http://bugs.icu-project.org/trac/ticket/4229
  4. http://bugs.icu-project.org/trac/ticket/3730
  5. FYI: http://www.sil.org/iso639-3/codes.asp?order=639_3&letter=f
  6. http://www.iana.org/assignments/language-subtag-registry 
  7. Agreed to fix bugs that crop up to move towards BCP 47 support
  1. "digit" locale keyword (Yoshito)
  1. Java Locale th_TH_TH designate Thai digit to be used for numbers
  2. No locale mapping to ICU
  3. No definition in the current CLDR - http://unicode.org/cldr/bugs/locale-bugs?findid=1471 
  4. Past discussion - digit=[native|alternative] or digit=<script>
  1. wintz.c (Eric)
  1. windows registry key: http://support.microsoft.com/kb/914387
  1. Milestone 2 (Steven)

2008/3/5 Agenda

Note:  Google people can not make the meeting today.  We can either postpone to next week, or schedule some other time to discuss specific issues.  -- Andy

This meeting was canceled.  -- Yoshito

2008/2/27 Agenda

  1. ISO-2022 and HZ non-shortest escape sequence (Peter)
  1. icu-project.org availability/upgrade (Michael)
  1. a

2008/2/20 Agenda

  1. ucol_getSortKey error handling (George)
  1. Changing the semantics of YEAR_WOY field? (Yoshito)

2008/2/13 Agenda

  1. Java SPI implementation packaging (Yoshito)
  1. DateFormatSymbols - getDateFormatBundle/initializeData (Yoshito/Doug)
  1. Coverity (George)
  2. Meeting time (Mark)
  1. Markus will be in Central European timezone for a while.
  2. Eric is in Hawaii.
  3. We may want to move the meeting earlier.
  1. Whitespace and syntax characters (Mark)

2008/2/6 Agenda

  1. ICU4C error behavior (e.g., UnicodeSet.isBogus) (Markus, George, Michael)
  1. Report from Unicode Symbols Subcommittee meetings during UTC (Markus)

2008/1/30 Agenda

  1. Confirm ICU 4.0 schedule after CLDR 1.6 discussion, set further intermediate dates (e.g., d01 & d02)
  1. Everyone agrees to postpone schedule by one month. (no disagreement)
  2. Milestone 2 postponed by 18 days.
  3. Milestone 2 reminder
  1. All tests should pass on reference platforms.
  2. Timebomb failures should be fixed.
  3. Good checkpoint to finish up reviews.

2008/1/23 Agenda

  1. ICU 4.0 schedule (Yoshito)
  1. Unicode 5.1 update status (Markus)
  1. Mutexes (Andy)

2008/1/16 Agenda

  1. ICU 4.0 schedule (Yoshito)
  1. Include CLDR 1.6 or not? CLDR vs. ICU schedules, not enough time.
  2. Tentatively move out ICU 4.0 release date to 2008-jul-02
  1. ICU features for new structure in CLDR 1.6 (Mark)
  1. Likely subtag structure from CLDR 1.5.1 supplemental data (Yoshito)

2008/1/9 Agenda

  1. Please review your tickets (George) http://bugs.icu-project.org/trac/query?status=new&status=assigned&status=reopened&group=revw&order=priority&revw=%21
  2. What is the schedule for Vladimir's and Doug's tickets? (George)
  1. Is the ICU4J charset API able to go draft in 4.0 release? (George)
  1. Backwards string search (Eric)
  1. Discuss more offline
  2. Eric to schedule a meeting with Andy, Vladimir, Mark, Michael, Lloyd and Claire.
  1. CLDR bugs. (Mark)
  1. Report on meeting from yesterday
  2. If there are items in ICU that depend on CLDR features, we want to get those in by Tuesday.
  3. Mark to forward note that he sent.
  1. Copyright scanner (Eric)
  1. Noticed that the copyright scanner sometimes didn't work.
  2. Eric to run copyright scanner to see what else made it through the scanner.

Past Meetings