Context Navigation

← Previous Revision
Latest Revision
Next Revision →
Blame
Revision Log

ncurses-intro.html@ 2645

Visit:

Last change on this file since 2645 was 2621, checked in by bird, 20 years ago
GNU ncurses 5.5
File size: 116.5 KB

Line
1	<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 3.0//EN">
2	<!--
3	$Id: ncurses-intro.html,v 1.40 2004/06/05 19:10:10 tom Exp $
4	-->
5	<HTML>
6	<HEAD>
7	<TITLE>Writing Programs with NCURSES</TITLE>
8	<link rev="made" href="mailto:bugs-ncurses@gnu.org">
9	<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
10	</HEAD>
11	<BODY>
12
13	<H1>Writing Programs with NCURSES</H1>
14
15	<BLOCKQUOTE>
16	by Eric S. Raymond and Zeyd M. Ben-Halim<BR>
17	updates since release 1.9.9e by Thomas Dickey
18	</BLOCKQUOTE>
19
20	<H1>Contents</H1>
21	<UL>
22	<LI><A HREF="#introduction">Introduction</A>
23	<UL>
24	<LI><A HREF="#history">A Brief History of Curses</A>
25	<LI><A HREF="#scope">Scope of This Document</A>
26	<LI><A HREF="#terminology">Terminology</A>
27	</UL>
28	<LI><A HREF="#curses">The Curses Library</A>
29	<UL>
30	<LI><A HREF="#overview">An Overview of Curses</A>
31	<UL>
32	<LI><A HREF="#compiling">Compiling Programs using Curses</A>
33	<LI><A HREF="#updating">Updating the Screen</A>
34	<LI><A HREF="#stdscr">Standard Windows and Function Naming Conventions</A>
35	<LI><A HREF="#variables">Variables</A>
36	</UL>
37	<LI><A HREF="#using">Using the Library</A>
38	<UL>
39	<LI><A HREF="#starting">Starting up</A>
40	<LI><A HREF="#output">Output</A>
41	<LI><A HREF="#input">Input</A>
42	<LI><A HREF="#formschars">Using Forms Characters</A>
43	<LI><A HREF="#attributes">Character Attributes and Color</A>
44	<LI><A HREF="#mouse">Mouse Interfacing</A>
45	<LI><A HREF="#finishing">Finishing Up</A>
46	</UL>
47	<LI><A HREF="#functions">Function Descriptions</A>
48	<UL>
49	<LI><A HREF="#init">Initialization and Wrapup</A>
50	<LI><A HREF="#flush">Causing Output to the Terminal</A>
51	<LI><A HREF="#lowlevel">Low-Level Capability Access</A>
52	<LI><A HREF="#debugging">Debugging</A>
53	</UL>
54	<LI><A HREF="#hints">Hints, Tips, and Tricks</A>
55	<UL>
56	<LI><A HREF="#caution">Some Notes of Caution</A>
57	<LI><A HREF="#leaving">Temporarily Leaving ncurses Mode</A>
58	<LI><A HREF="#xterm">Using <CODE>ncurses</CODE> under <CODE>xterm</CODE></A>
59	<LI><A HREF="#screens">Handling Multiple Terminal Screens</A>
60	<LI><A HREF="#testing">Testing for Terminal Capabilities</A>
61	<LI><A HREF="#tuning">Tuning for Speed</A>
62	<LI><A HREF="#special">Special Features of <CODE>ncurses</CODE></A>
63	</UL>
64	<LI><A HREF="#compat">Compatibility with Older Versions</A>
65	<UL>
66	<LI><A HREF="#refbug">Refresh of Overlapping Windows</A>
67	<LI><A HREF="#backbug">Background Erase</A>
68	</UL>
69	<LI><A HREF="#xsifuncs">XSI Curses Conformance</A>
70	</UL>
71	<LI><A HREF="#panels">The Panels Library</A>
72	<UL>
73	<LI><A HREF="#pcompile">Compiling With the Panels Library</A>
74	<LI><A HREF="#poverview">Overview of Panels</A>
75	<LI><A HREF="#pstdscr">Panels, Input, and the Standard Screen</A>
76	<LI><A HREF="#hiding">Hiding Panels</A>
77	<LI><A HREF="#pmisc">Miscellaneous Other Facilities</A>
78	</UL>
79	<LI><A HREF="#menu">The Menu Library</A>
80	<UL>
81	<LI><A HREF="#mcompile">Compiling with the menu Library</A>
82	<LI><A HREF="#moverview">Overview of Menus</A>
83	<LI><A HREF="#mselect">Selecting items</A>
84	<LI><A HREF="#mdisplay">Menu Display</A>
85	<LI><A HREF="#mwindows">Menu Windows</A>
86	<LI><A HREF="#minput">Processing Menu Input</A>
87	<LI><A HREF="#mmisc">Miscellaneous Other Features</A>
88	</UL>
89	<LI><A HREF="#form">The Forms Library</A>
90	<UL>
91	<LI><A HREF="#fcompile">Compiling with the forms Library</A>
92	<LI><A HREF="#foverview">Overview of Forms</A>
93	<LI><A HREF="#fcreate">Creating and Freeing Fields and Forms</A>
94	<LI><A HREF="#fattributes">Fetching and Changing Field Attributes</A>
95	<UL>
96	<LI><A HREF="#fsizes">Fetching Size and Location Data</A>
97	<LI><A HREF="#flocation">Changing the Field Location</A>
98	<LI><A HREF="#fjust">The Justification Attribute</A>
99	<LI><A HREF="#fdispatts">Field Display Attributes</A>
100	<LI><A HREF="#foptions">Field Option Bits</A>
101	<LI><A HREF="#fstatus">Field Status</A>
102	<LI><A HREF="#fuser">Field User Pointer</A>
103	</UL>
104	<LI><A HREF="#fdynamic">Variable-Sized Fields</A>
105	<LI><A HREF="#fvalidation">Field Validation</A>
106	<UL>
107	<LI><A HREF="#ftype_alpha">TYPE_ALPHA</A>
108	<LI><A HREF="#ftype_alnum">TYPE_ALNUM</A>
109	<LI><A HREF="#ftype_enum">TYPE_ENUM</A>
110	<LI><A HREF="#ftype_integer">TYPE_INTEGER</A>
111	<LI><A HREF="#ftype_numeric">TYPE_NUMERIC</A>
112	<LI><A HREF="#ftype_regexp">TYPE_REGEXP</A>
113	</UL>
114	<LI><A HREF="#fbuffer">Direct Field Buffer Manipulation</A>
115	<LI><A HREF="#formattrs">Attributes of Forms</A>
116	<LI><A HREF="#fdisplay">Control of Form Display</A>
117	<LI><A HREF="#fdriver">Input Processing in the Forms Driver</A>
118	<UL>
119	<LI><A HREF="#fpage">Page Navigation Requests</A>
120	<LI><A HREF="#ffield">Inter-Field Navigation Requests</A>
121	<LI><A HREF="#fifield">Intra-Field Navigation Requests</A>
122	<LI><A HREF="#fscroll">Scrolling Requests</A>
123	<LI><A HREF="#fedit">Field Editing Requests</A>
124	<LI><A HREF="#forder">Order Requests</A>
125	<LI><A HREF="#fappcmds">Application Commands</A>
126	</UL>
127	<LI><A HREF="#fhooks">Field Change Hooks</A>
128	<LI><A HREF="#ffocus">Field Change Commands</A>
129	<LI><A HREF="#frmoptions">Form Options</A>
130	<LI><A HREF="#fcustom">Custom Validation Types</A>
131	<UL>
132	<LI><A HREF="#flinktypes">Union Types</A>
133	<LI><A HREF="#fnewtypes">New Field Types</A>
134	<LI><A HREF="#fcheckargs">Validation Function Arguments</A>
135	<LI><A HREF="#fcustorder">Order Functions For Custom Types</A>
136	<LI><A HREF="#fcustprobs">Avoiding Problems</A>
137	</UL>
138	</UL>
139	</UL>
140
141	<HR>
142	<H1><A NAME="introduction">Introduction</A></H1>
143
144	This document is an introduction to programming with <CODE>curses</CODE>. It is
145	not an exhaustive reference for the curses Application Programming Interface
146	(API); that role is filled by the <CODE>curses</CODE> manual pages. Rather, it
147	is intended to help C programmers ease into using the package. <P>
148
149	This document is aimed at C applications programmers not yet specifically
150	familiar with ncurses. If you are already an experienced <CODE>curses</CODE>
151	programmer, you should nevertheless read the sections on
152	<A HREF="#mouse">Mouse Interfacing</A>, <A HREF="#debugging">Debugging</A>,
153	<A HREF="#compat">Compatibility with Older Versions</A>,
154	and <A HREF="#hints">Hints, Tips, and Tricks</A>. These will bring you up
155	to speed on the special features and quirks of the <CODE>ncurses</CODE>
156	implementation. If you are not so experienced, keep reading. <P>
157
158	The <CODE>curses</CODE> package is a subroutine library for
159	terminal-independent screen-painting and input-event handling which
160	presents a high level screen model to the programmer, hiding differences
161	between terminal types and doing automatic optimization of output to change
162	one screen full of text into another. <CODE>Curses</CODE> uses terminfo, which
163	is a database format that can describe the capabilities of thousands of
164	different terminals. <P>
165
166	The <CODE>curses</CODE> API may seem something of an archaism on UNIX desktops
167	increasingly dominated by X, Motif, and Tcl/Tk. Nevertheless, UNIX still
168	supports tty lines and X supports <EM>xterm(1)</EM>; the <CODE>curses</CODE>
169	API has the advantage of (a) back-portability to character-cell terminals,
170	and (b) simplicity. For an application that does not require bit-mapped
171	graphics and multiple fonts, an interface implementation using <CODE>curses</CODE>
172	will typically be a great deal simpler and less expensive than one using an
173	X toolkit.
174
175	<H2><A NAME="history">A Brief History of Curses</A></H2>
176
177	Historically, the first ancestor of <CODE>curses</CODE> was the routines written to
178	provide screen-handling for the game <CODE>rogue</CODE>; these used the
179	already-existing <CODE>termcap</CODE> database facility for describing terminal
180	capabilities. These routines were abstracted into a documented library and
181	first released with the early BSD UNIX versions. <P>
182
183	System III UNIX from Bell Labs featured a rewritten and much-improved
184	<CODE>curses</CODE> library. It introduced the terminfo format. Terminfo is based
185	on Berkeley's termcap database, but contains a number of improvements and
186	extensions. Parameterized capabilities strings were introduced, making it
187	possible to describe multiple video attributes, and colors and to handle far
188	more unusual terminals than possible with termcap. In the later AT&T
189	System V releases, <CODE>curses</CODE> evolved to use more facilities and offer
190	more capabilities, going far beyond BSD curses in power and flexibility.
191
192	<H2><A NAME="scope">Scope of This Document</A></H2>
193
194	This document describes <CODE>ncurses</CODE>, a free implementation of
195	the System V <CODE>curses</CODE> API with some clearly marked extensions.
196	It includes the following System V curses features:
197	<UL>
198	<LI>Support for multiple screen highlights (BSD curses could only
199	handle one `standout' highlight, usually reverse-video).
200	<LI>Support for line- and box-drawing using forms characters.
201	<LI>Recognition of function keys on input.
202	<LI>Color support.
203	<LI>Support for pads (windows of larger than screen size on which the
204	screen or a subwindow defines a viewport).
205	</UL>
206
207	Also, this package makes use of the insert and delete line and character
208	features of terminals so equipped, and determines how to optimally use these
209	features with no help from the programmer. It allows arbitrary combinations of
210	video attributes to be displayed, even on terminals that leave ``magic
211	cookies'' on the screen to mark changes in attributes. <P>
212
213	The <CODE>ncurses</CODE> package can also capture and use event reports from a
214	mouse in some environments (notably, xterm under the X window system). This
215	document includes tips for using the mouse. <P>
216
217	The <CODE>ncurses</CODE> package was originated by Pavel Curtis. The original
218	maintainer of this package is
219	<A HREF="mailto:zmbenhal@netcom.com">Zeyd Ben-Halim</A>
220	<zmbenhal@netcom.com>.
221	<A HREF="mailto:esr@snark.thyrsus.com">Eric S. Raymond</A>
222	<esr@snark.thyrsus.com>
223	wrote many of the new features in versions after 1.8.1
224	and wrote most of this introduction.
225	Jürgen Pfeifer
226	wrote all of the menu and forms code as well as the
227	<A HREF="http://www.adahome.com">Ada95</A> binding.
228	Ongoing work is being done by
229	<A HREF="mailto:dickey@invisible-island.net">Thomas Dickey</A> (maintainer).
230	Contact the current maintainers at
231	<A HREF="mailto:bug-ncurses@gnu.org">bug-ncurses@gnu.org</A>.
232	<P>
233
234	This document also describes the <A HREF="#panels">panels</A> extension library,
235	similarly modeled on the SVr4 panels facility. This library allows you to
236	associate backing store with each of a stack or deck of overlapping windows,
237	and provides operations for moving windows around in the stack that change
238	their visibility in the natural way (handling window overlaps). <P>
239
240	Finally, this document describes in detail the <A HREF="#menu">menus</A> and <A
241	HREF="#form">forms</A> extension libraries, also cloned from System V,
242	which support easy construction and sequences of menus and fill-in
243	forms.
244
245
246	<H2><A NAME="terminology">Terminology</A></H2>
247
248	In this document, the following terminology is used with reasonable
249	consistency:
250
251	<DL>
252	<DT> window
253	<DD>
254	A data structure describing a sub-rectangle of the screen (possibly the
255	entire screen). You can write to a window as though it were a miniature
256	screen, scrolling independently of other windows on the physical screen.
257	<DT> screens
258	<DD>
259	A subset of windows which are as large as the terminal screen, i.e., they start
260	at the upper left hand corner and encompass the lower right hand corner. One
261	of these, <CODE>stdscr</CODE>, is automatically provided for the programmer.
262	<DT> terminal screen
263	<DD>
264	The package's idea of what the terminal display currently looks like, i.e.,
265	what the user sees now. This is a special screen.
266	</DL>
267
268	<H1><A NAME="curses">The Curses Library</A></H1>
269
270	<H2><A NAME="overview">An Overview of Curses</A></H2>
271
272	<H3><A NAME="compiling">Compiling Programs using Curses</A></H3>
273
274	In order to use the library, it is necessary to have certain types and
275	variables defined. Therefore, the programmer must have a line:
276
277	<PRE>
278	#include <curses.h>
279	</PRE>
280
281	at the top of the program source. The screen package uses the Standard I/O
282	library, so <CODE><curses.h></CODE> includes
283	<CODE><stdio.h></CODE>. <CODE><curses.h></CODE> also includes
284	<CODE><termios.h></CODE>, <CODE><termio.h></CODE>, or
285	<CODE><sgtty.h></CODE> depending on your system. It is redundant (but
286	harmless) for the programmer to do these includes, too. In linking with
287	<CODE>curses</CODE> you need to have <CODE>-lncurses</CODE> in your LDFLAGS or on the
288	command line. There is no need for any other libraries.
289
290	<H3><A NAME="updating">Updating the Screen</A></H3>
291
292	In order to update the screen optimally, it is necessary for the routines to
293	know what the screen currently looks like and what the programmer wants it to
294	look like next. For this purpose, a data type (structure) named WINDOW is
295	defined which describes a window image to the routines, including its starting
296	position on the screen (the (y, x) coordinates of the upper left hand corner)
297	and its size. One of these (called <CODE>curscr</CODE>, for current screen) is a
298	screen image of what the terminal currently looks like. Another screen (called
299	<CODE>stdscr</CODE>, for standard screen) is provided by default to make changes
300	on. <P>
301
302	A window is a purely internal representation. It is used to build and store a
303	potential image of a portion of the terminal. It doesn't bear any necessary
304	relation to what is really on the terminal screen; it's more like a
305	scratchpad or write buffer. <P>
306
307	To make the section of physical screen corresponding to a window reflect the
308	contents of the window structure, the routine <CODE>refresh()</CODE> (or
309	<CODE>wrefresh()</CODE> if the window is not <CODE>stdscr</CODE>) is called. <P>
310
311	A given physical screen section may be within the scope of any number of
312	overlapping windows. Also, changes can be made to windows in any order,
313	without regard to motion efficiency. Then, at will, the programmer can
314	effectively say ``make it look like this,'' and let the package implementation
315	determine the most efficient way to repaint the screen.
316
317	<H3><A NAME="stdscr">Standard Windows and Function Naming Conventions</A></H3>
318
319	As hinted above, the routines can use several windows, but two are
320	automatically given: <CODE>curscr</CODE>, which knows what the terminal looks like,
321	and <CODE>stdscr</CODE>, which is what the programmer wants the terminal to look
322	like next. The user should never actually access <CODE>curscr</CODE> directly.
323	Changes should be made to through the API, and then the routine
324	<CODE>refresh()</CODE> (or <CODE>wrefresh()</CODE>) called. <P>
325
326	Many functions are defined to use <CODE>stdscr</CODE> as a default screen. For
327	example, to add a character to <CODE>stdscr</CODE>, one calls <CODE>addch()</CODE> with
328	the desired character as argument. To write to a different window. use the
329	routine <CODE>waddch()</CODE> (for `w'indow-specific addch()) is provided. This
330	convention of prepending function names with a `w' when they are to be
331	applied to specific windows is consistent. The only routines which do not
332	follow it are those for which a window must always be specified. <P>
333
334	In order to move the current (y, x) coordinates from one point to another, the
335	routines <CODE>move()</CODE> and <CODE>wmove()</CODE> are provided. However, it is
336	often desirable to first move and then perform some I/O operation. In order to
337	avoid clumsiness, most I/O routines can be preceded by the prefix 'mv' and
338	the desired (y, x) coordinates prepended to the arguments to the function. For
339	example, the calls
340
341	<PRE>
342	move(y, x);
343	addch(ch);
344	</PRE>
345
346	can be replaced by
347
348	<PRE>
349	mvaddch(y, x, ch);
350	</PRE>
351
352	and
353
354	<PRE>
355	wmove(win, y, x);
356	waddch(win, ch);
357	</PRE>
358
359	can be replaced by
360
361	<PRE>
362	mvwaddch(win, y, x, ch);
363	</PRE>
364
365	Note that the window description pointer (win) comes before the added (y, x)
366	coordinates. If a function requires a window pointer, it is always the first
367	parameter passed.
368
369	<H3><A NAME="variables">Variables</A></H3>
370
371	The <CODE>curses</CODE> library sets some variables describing the terminal
372	capabilities.
373
374	<PRE>
375	type name description
376	------------------------------------------------------------------
377	int LINES number of lines on the terminal
378	int COLS number of columns on the terminal
379	</PRE>
380
381	The <CODE>curses.h</CODE> also introduces some <CODE>#define</CODE> constants and types
382	of general usefulness:
383
384	<DL>
385	<DT> <CODE>bool</CODE>
386	<DD> boolean type, actually a `char' (e.g., <CODE>bool doneit;</CODE>)
387	<DT> <CODE>TRUE</CODE>
388	<DD> boolean `true' flag (1).
389	<DT> <CODE>FALSE</CODE>
390	<DD> boolean `false' flag (0).
391	<DT> <CODE>ERR</CODE>
392	<DD> error flag returned by routines on a failure (-1).
393	<DT> <CODE>OK</CODE>
394	<DD> error flag returned by routines when things go right.
395	</DL>
396
397	<H2><A NAME="using">Using the Library</A></H2>
398
399	Now we describe how to actually use the screen package. In it, we assume all
400	updating, reading, etc. is applied to <CODE>stdscr</CODE>. These instructions will
401	work on any window, providing you change the function names and parameters as
402	mentioned above. <P>
403
404	Here is a sample program to motivate the discussion:
405
406	<PRE>
407	#include <curses.h>
408	#include <signal.h>
409
410	static void finish(int sig);
411
412	int
413	main(int argc, char *argv[])
414	{
415	int num = 0;
416
417	/* initialize your non-curses data structures here */
418
419	(void) signal(SIGINT, finish); /* arrange interrupts to terminate */
420
421	(void) initscr(); /* initialize the curses library */
422	keypad(stdscr, TRUE); /* enable keyboard mapping */
423	(void) nonl(); /* tell curses not to do NL->CR/NL on output */
424	(void) cbreak(); /* take input chars one at a time, no wait for \n */
425	(void) echo(); /* echo input - in color */
426
427	if (has_colors())
428	{
429	start_color();
430
431	/*
432	* Simple color assignment, often all we need. Color pair 0 cannot
433	* be redefined. This example uses the same value for the color
434	* pair as for the foreground color, though of course that is not
435	* necessary:
436	*/
437	init_pair(1, COLOR_RED, COLOR_BLACK);
438	init_pair(2, COLOR_GREEN, COLOR_BLACK);
439	init_pair(3, COLOR_YELLOW, COLOR_BLACK);
440	init_pair(4, COLOR_BLUE, COLOR_BLACK);
441	init_pair(5, COLOR_CYAN, COLOR_BLACK);
442	init_pair(6, COLOR_MAGENTA, COLOR_BLACK);
443	init_pair(7, COLOR_WHITE, COLOR_BLACK);
444	}
445
446	for (;;)
447	{
448	int c = getch(); /* refresh, accept single keystroke of input */
449	attrset(COLOR_PAIR(num % 8));
450	num++;
451
452	/* process the command keystroke */
453	}
454
455	finish(0); /* we're done */
456	}
457
458	static void finish(int sig)
459	{
460	endwin();
461
462	/* do your non-curses wrapup here */
463
464	exit(0);
465	}
466	</PRE>
467
468	<H3><A NAME="starting">Starting up</A></H3>
469
470	In order to use the screen package, the routines must know about terminal
471	characteristics, and the space for <CODE>curscr</CODE> and <CODE>stdscr</CODE> must be
472	allocated. These function <CODE>initscr()</CODE> does both these things. Since it
473	must allocate space for the windows, it can overflow memory when attempting to
474	do so. On the rare occasions this happens, <CODE>initscr()</CODE> will terminate
475	the program with an error message. <CODE>initscr()</CODE> must always be called
476	before any of the routines which affect windows are used. If it is not, the
477	program will core dump as soon as either <CODE>curscr</CODE> or <CODE>stdscr</CODE> are
478	referenced. However, it is usually best to wait to call it until after you are
479	sure you will need it, like after checking for startup errors. Terminal status
480	changing routines like <CODE>nl()</CODE> and <CODE>cbreak()</CODE> should be called
481	after <CODE>initscr()</CODE>. <P>
482
483	Once the screen windows have been allocated, you can set them up for
484	your program. If you want to, say, allow a screen to scroll, use
485	<CODE>scrollok()</CODE>. If you want the cursor to be left in place after
486	the last change, use <CODE>leaveok()</CODE>. If this isn't done,
487	<CODE>refresh()</CODE> will move the cursor to the window's current (y, x)
488	coordinates after updating it. <P>
489
490	You can create new windows of your own using the functions <CODE>newwin()</CODE>,
491	<CODE>derwin()</CODE>, and <CODE>subwin()</CODE>. The routine <CODE>delwin()</CODE> will
492	allow you to get rid of old windows. All the options described above can be
493	applied to any window.
494
495	<H3><A NAME="output">Output</A></H3>
496
497	Now that we have set things up, we will want to actually update the terminal.
498	The basic functions used to change what will go on a window are
499	<CODE>addch()</CODE> and <CODE>move()</CODE>. <CODE>addch()</CODE> adds a character at the
500	current (y, x) coordinates. <CODE>move()</CODE> changes the current (y, x)
501	coordinates to whatever you want them to be. It returns <CODE>ERR</CODE> if you
502	try to move off the window. As mentioned above, you can combine the two into
503	<CODE>mvaddch()</CODE> to do both things at once. <P>
504
505	The other output functions, such as <CODE>addstr()</CODE> and <CODE>printw()</CODE>,
506	all call <CODE>addch()</CODE> to add characters to the window. <P>
507
508	After you have put on the window what you want there, when you want the portion
509	of the terminal covered by the window to be made to look like it, you must call
510	<CODE>refresh()</CODE>. In order to optimize finding changes, <CODE>refresh()</CODE>
511	assumes that any part of the window not changed since the last
512	<CODE>refresh()</CODE> of that window has not been changed on the terminal, i.e.,
513	that you have not refreshed a portion of the terminal with an overlapping
514	window. If this is not the case, the routine <CODE>touchwin()</CODE> is provided
515	to make it look like the entire window has been changed, thus making
516	<CODE>refresh()</CODE> check the whole subsection of the terminal for changes. <P>
517
518	If you call <CODE>wrefresh()</CODE> with <CODE>curscr</CODE> as its argument, it will
519	make the screen look like <CODE>curscr</CODE> thinks it looks like. This is useful
520	for implementing a command which would redraw the screen in case it get messed
521	up.
522
523	<H3><A NAME="input">Input</A></H3>
524
525	The complementary function to <CODE>addch()</CODE> is <CODE>getch()</CODE> which, if
526	echo is set, will call <CODE>addch()</CODE> to echo the character. Since the
527	screen package needs to know what is on the terminal at all times, if
528	characters are to be echoed, the tty must be in raw or cbreak mode. Since
529	initially the terminal has echoing enabled and is in ordinary ``cooked'' mode,
530	one or the other has to changed before calling <CODE>getch()</CODE>; otherwise,
531	the program's output will be unpredictable. <P>
532
533	When you need to accept line-oriented input in a window, the functions
534	<CODE>wgetstr()</CODE> and friends are available. There is even a <CODE>wscanw()</CODE>
535	function that can do <CODE>scanf()</CODE>(3)-style multi-field parsing on window
536	input. These pseudo-line-oriented functions turn on echoing while they
537	execute. <P>
538
539	The example code above uses the call <CODE>keypad(stdscr, TRUE)</CODE> to enable
540	support for function-key mapping. With this feature, the <CODE>getch()</CODE> code
541	watches the input stream for character sequences that correspond to arrow and
542	function keys. These sequences are returned as pseudo-character values. The
543	<CODE>#define</CODE> values returned are listed in the <CODE>curses.h</CODE> The
544	mapping from sequences to <CODE>#define</CODE> values is determined by
545	<CODE>key_</CODE> capabilities in the terminal's terminfo entry.
546
547	<H3><A NAME="formschars">Using Forms Characters</A></H3>
548
549	The <CODE>addch()</CODE> function (and some others, including <CODE>box()</CODE> and
550	<CODE>border()</CODE>) can accept some pseudo-character arguments which are specially
551	defined by <CODE>ncurses</CODE>. These are <CODE>#define</CODE> values set up in
552	the <CODE>curses.h</CODE> header; see there for a complete list (look for
553	the prefix <CODE>ACS_</CODE>). <P>
554
555	The most useful of the ACS defines are the forms-drawing characters. You can
556	use these to draw boxes and simple graphs on the screen. If the terminal
557	does not have such characters, <CODE>curses.h</CODE> will map them to a
558	recognizable (though ugly) set of ASCII defaults.
559
560	<H3><A NAME="attributes">Character Attributes and Color</A></H3>
561
562	The <CODE>ncurses</CODE> package supports screen highlights including standout,
563	reverse-video, underline, and blink. It also supports color, which is treated
564	as another kind of highlight. <P>
565
566	Highlights are encoded, internally, as high bits of the pseudo-character type
567	(<CODE>chtype</CODE>) that <CODE>curses.h</CODE> uses to represent the contents of a
568	screen cell. See the <CODE>curses.h</CODE> header file for a complete list of
569	highlight mask values (look for the prefix <CODE>A_</CODE>).<P>
570
571	There are two ways to make highlights. One is to logical-or the value of the
572	highlights you want into the character argument of an <CODE>addch()</CODE> call,
573	or any other output call that takes a <CODE>chtype</CODE> argument. <P>
574
575	The other is to set the current-highlight value. This is logical-or'ed with
576	any highlight you specify the first way. You do this with the functions
577	<CODE>attron()</CODE>, <CODE>attroff()</CODE>, and <CODE>attrset()</CODE>; see the manual
578	pages for details.
579
580	Color is a special kind of highlight. The package actually thinks in terms
581	of color pairs, combinations of foreground and background colors. The sample
582	code above sets up eight color pairs, all of the guaranteed-available colors
583	on black. Note that each color pair is, in effect, given the name of its
584	foreground color. Any other range of eight non-conflicting values could
585	have been used as the first arguments of the <CODE>init_pair()</CODE> values. <P>
586
587	Once you've done an <CODE>init_pair()</CODE> that creates color-pair N, you can
588	use <CODE>COLOR_PAIR(N)</CODE> as a highlight that invokes that particular
589	color combination. Note that <CODE>COLOR_PAIR(N)</CODE>, for constant N,
590	is itself a compile-time constant and can be used in initializers.
591
592	<H3><A NAME="mouse">Mouse Interfacing</A></H3>
593
594	The <CODE>ncurses</CODE> library also provides a mouse interface.
595	<!-- The 'note' tag is not portable enough -->
596	<blockquote>
597	<strong>NOTE:</strong> this facility is specific to <CODE>ncurses</CODE>, it is not part of either
598	the XSI Curses standard, nor of System V Release 4, nor BSD curses.
599	System V Release 4 curses contains code with similar interface definitions,
600	however it is not documented. Other than by disassembling the library, we
601	have no way to determine exactly how that mouse code works.
602	Thus, we recommend that you wrap mouse-related code in an #ifdef using the
603	feature macro NCURSES_MOUSE_VERSION so it will not be compiled and linked
604	on non-ncurses systems.
605	</blockquote>
606
607	Presently, mouse event reporting works in the following environments:
608	<ul>
609	<li>xterm and similar programs such as rxvt.
610	<li>Linux console, when configured with <CODE>gpm</CODE>(1), Alessandro
611	Rubini's mouse server.
612	<li>FreeBSD sysmouse (console)
613	<li>OS/2 EMX
614	</ul>
615	<P>
616	The mouse interface is very simple. To activate it, you use the function
617	<CODE>mousemask()</CODE>, passing it as first argument a bit-mask that specifies
618	what kinds of events you want your program to be able to see. It will
619	return the bit-mask of events that actually become visible, which may differ
620	from the argument if the mouse device is not capable of reporting some of
621	the event types you specify. <P>
622
623	Once the mouse is active, your application's command loop should watch
624	for a return value of <CODE>KEY_MOUSE</CODE> from <CODE>wgetch()</CODE>. When
625	you see this, a mouse event report has been queued. To pick it off
626	the queue, use the function <CODE>getmouse()</CODE> (you must do this before
627	the next <CODE>wgetch()</CODE>, otherwise another mouse event might come
628	in and make the first one inaccessible). <P>
629
630	Each call to <CODE>getmouse()</CODE> fills a structure (the address of which you'll
631	pass it) with mouse event data. The event data includes zero-origin,
632	screen-relative character-cell coordinates of the mouse pointer. It also
633	includes an event mask. Bits in this mask will be set, corresponding
634	to the event type being reported. <P>
635
636	The mouse structure contains two additional fields which may be
637	significant in the future as ncurses interfaces to new kinds of
638	pointing device. In addition to x and y coordinates, there is a slot
639	for a z coordinate; this might be useful with touch-screens that can
640	return a pressure or duration parameter. There is also a device ID
641	field, which could be used to distinguish between multiple pointing
642	devices. <P>
643
644	The class of visible events may be changed at any time via <CODE>mousemask()</CODE>.
645	Events that can be reported include presses, releases, single-, double- and
646	triple-clicks (you can set the maximum button-down time for clicks). If
647	you don't make clicks visible, they will be reported as press-release
648	pairs. In some environments, the event mask may include bits reporting
649	the state of shift, alt, and ctrl keys on the keyboard during the event. <P>
650
651	A function to check whether a mouse event fell within a given window is
652	also supplied. You can use this to see whether a given window should
653	consider a mouse event relevant to it. <P>
654
655	Because mouse event reporting will not be available in all
656	environments, it would be unwise to build <CODE>ncurses</CODE>
657	applications that <EM>require</EM> the use of a mouse. Rather, you should
658	use the mouse as a shortcut for point-and-shoot commands your application
659	would normally accept from the keyboard. Two of the test games in the
660	<CODE>ncurses</CODE> distribution (<CODE>bs</CODE> and <CODE>knight</CODE>) contain
661	code that illustrates how this can be done. <P>
662
663	See the manual page <CODE>curs_mouse(3X)</CODE> for full details of the
664	mouse-interface functions.
665
666	<H3><A NAME="finishing">Finishing Up</A></H3>
667
668	In order to clean up after the <CODE>ncurses</CODE> routines, the routine
669	<CODE>endwin()</CODE> is provided. It restores tty modes to what they were when
670	<CODE>initscr()</CODE> was first called, and moves the cursor down to the
671	lower-left corner. Thus, anytime after the call to initscr, <CODE>endwin()</CODE>
672	should be called before exiting.
673
674	<H2><A NAME="functions">Function Descriptions</A></H2>
675
676	We describe the detailed behavior of some important curses functions here, as a
677	supplement to the manual page descriptions.
678
679	<H3><A NAME="init">Initialization and Wrapup</A></H3>
680
681	<DL>
682	<DT> <CODE>initscr()</CODE>
683	<DD> The first function called should almost always be <CODE>initscr()</CODE>.
684	This will determine the terminal type and
685	initialize curses data structures. <CODE>initscr()</CODE> also arranges that
686	the first call to <CODE>refresh()</CODE> will clear the screen. If an error
687	occurs a message is written to standard error and the program
688	exits. Otherwise it returns a pointer to stdscr. A few functions may be
689	called before initscr (<CODE>slk_init()</CODE>, <CODE>filter()</CODE>,
690	<CODE>ripoffline()</CODE>, <CODE>use_env()</CODE>, and, if you are using multiple
691	terminals, <CODE>newterm()</CODE>.)
692	<DT> <CODE>endwin()</CODE>
693	<DD> Your program should always call <CODE>endwin()</CODE> before exiting or
694	shelling out of the program. This function will restore tty modes,
695	move the cursor to the lower left corner of the screen, reset the
696	terminal into the proper non-visual mode. Calling <CODE>refresh()</CODE>
697	or <CODE>doupdate()</CODE> after a temporary escape from the program will
698	restore the ncurses screen from before the escape.
699	<DT> <CODE>newterm(type, ofp, ifp)</CODE>
700	<DD> A program which outputs to more than one terminal should use
701	<CODE>newterm()</CODE> instead of <CODE>initscr()</CODE>. <CODE>newterm()</CODE> should
702	be called once for each terminal. It returns a variable of type
703	<CODE>SCREEN *</CODE> which should be saved as a reference to that
704	terminal.
705	(NOTE: a SCREEN variable is not a <em>screen</em> in the sense we
706	are describing in this introduction, but a collection of
707	parameters used to assist in optimizing the display.)
708	The arguments are the type of the terminal (a string) and
709	<CODE>FILE</CODE> pointers for the output and input of the terminal. If
710	type is NULL then the environment variable <CODE>$TERM</CODE> is used.
711	<CODE>endwin()</CODE> should called once at wrapup time for each terminal
712	opened using this function.
713	<DT> <CODE>set_term(new)</CODE>
714	<DD> This function is used to switch to a different terminal previously
715	opened by <CODE>newterm()</CODE>. The screen reference for the new terminal
716	is passed as the parameter. The previous terminal is returned by the
717	function. All other calls affect only the current terminal.
718	<DT> <CODE>delscreen(sp)</CODE>
719	<DD> The inverse of <CODE>newterm()</CODE>; deallocates the data structures
720	associated with a given <CODE>SCREEN</CODE> reference.
721	</DL>
722
723	<H3><A NAME="flush">Causing Output to the Terminal</A></H3>
724
725	<DL>
726	<DT> <CODE>refresh()</CODE> and <CODE>wrefresh(win)</CODE>
727	<DD> These functions must be called to actually get any output on
728	the terminal, as other routines merely manipulate data
729	structures. <CODE>wrefresh()</CODE> copies the named window to the physical
730	terminal screen, taking into account what is already
731	there in order to do optimizations. <CODE>refresh()</CODE> does a
732	refresh of <CODE>stdscr</CODE>. Unless <CODE>leaveok()</CODE> has been
733	enabled, the physical cursor of the terminal is left at the
734	location of the window's cursor.
735	<DT> <CODE>doupdate()</CODE> and <CODE>wnoutrefresh(win)</CODE>
736	<DD> These two functions allow multiple updates with more efficiency
737	than wrefresh. To use them, it is important to understand how curses
738	works. In addition to all the window structures, curses keeps two
739	data structures representing the terminal screen: a physical screen,
740	describing what is actually on the screen, and a virtual screen,
741	describing what the programmer wants to have on the screen. wrefresh
742	works by first copying the named window to the virtual screen
743	(<CODE>wnoutrefresh()</CODE>), and then calling the routine to update the
744	screen (<CODE>doupdate()</CODE>). If the programmer wishes to output
745	several windows at once, a series of calls to <CODE>wrefresh</CODE> will result
746	in alternating calls to <CODE>wnoutrefresh()</CODE> and <CODE>doupdate()</CODE>,
747	causing several bursts of output to the screen. By calling
748	<CODE>wnoutrefresh()</CODE> for each window, it is then possible to call
749	<CODE>doupdate()</CODE> once, resulting in only one burst of output, with
750	fewer total characters transmitted (this also avoids a visually annoying
751	flicker at each update).
752	</DL>
753
754	<H3><A NAME="lowlevel">Low-Level Capability Access</A></H3>
755
756	<DL>
757	<DT> <CODE>setupterm(term, filenum, errret)</CODE>
758	<DD> This routine is called to initialize a terminal's description, without setting
759	up the curses screen structures or changing the tty-driver mode bits.
760	<CODE>term</CODE> is the character string representing the name of the terminal
761	being used. <CODE>filenum</CODE> is the UNIX file descriptor of the terminal to
762	be used for output. <CODE>errret</CODE> is a pointer to an integer, in which a
763	success or failure indication is returned. The values returned can be 1 (all
764	is well), 0 (no such terminal), or -1 (some problem locating the terminfo
765	database). <P>
766
767	The value of <CODE>term</CODE> can be given as NULL, which will cause the value of
768	<CODE>TERM</CODE> in the environment to be used. The <CODE>errret</CODE> pointer can
769	also be given as NULL, meaning no error code is wanted. If <CODE>errret</CODE> is
770	defaulted, and something goes wrong, <CODE>setupterm()</CODE> will print an
771	appropriate error message and exit, rather than returning. Thus, a simple
772	program can call setupterm(0, 1, 0) and not worry about initialization
773	errors. <P>
774
775	After the call to <CODE>setupterm()</CODE>, the global variable <CODE>cur_term</CODE> is
776	set to point to the current structure of terminal capabilities. By calling
777	<CODE>setupterm()</CODE> for each terminal, and saving and restoring
778	<CODE>cur_term</CODE>, it is possible for a program to use two or more terminals at
779	once. <CODE>Setupterm()</CODE> also stores the names section of the terminal
780	description in the global character array <CODE>ttytype[]</CODE>. Subsequent calls
781	to <CODE>setupterm()</CODE> will overwrite this array, so you'll have to save it
782	yourself if need be.
783	</DL>
784
785	<H3><A NAME="debugging">Debugging</A></H3>
786
787	<!-- The 'note' tag is not portable enough -->
788	<blockquote>
789	<strong>NOTE:</strong> These functions are not part of the standard curses API!
790	</blockquote>
791
792	<DL>
793	<DT> <CODE>trace()</CODE>
794	<DD>
795	This function can be used to explicitly set a trace level. If the
796	trace level is nonzero, execution of your program will generate a file
797	called `trace' in the current working directory containing a report on
798	the library's actions. Higher trace levels enable more detailed (and
799	verbose) reporting -- see comments attached to <CODE>TRACE_</CODE> defines
800	in the <CODE>curses.h</CODE> file for details. (It is also possible to set
801	a trace level by assigning a trace level value to the environment variable
802	<CODE>NCURSES_TRACE</CODE>).
803	<DT> <CODE>_tracef()</CODE>
804	<DD>
805	This function can be used to output your own debugging information. It is only
806	available only if you link with -lncurses_g. It can be used the same way as
807	<CODE>printf()</CODE>, only it outputs a newline after the end of arguments.
808	The output goes to a file called <CODE>trace</CODE> in the current directory.
809	</DL>
810
811	Trace logs can be difficult to interpret due to the sheer volume of
812	data dumped in them. There is a script called <STRONG>tracemunch</STRONG>
813	included with the <CODE>ncurses</CODE> distribution that can alleviate
814	this problem somewhat; it compacts long sequences of similar operations into
815	more succinct single-line pseudo-operations. These pseudo-ops can be
816	distinguished by the fact that they are named in capital letters.
817
818	<H2><A NAME="hints">Hints, Tips, and Tricks</A></H2>
819
820	The <CODE>ncurses</CODE> manual pages are a complete reference for this library.
821	In the remainder of this document, we discuss various useful methods that
822	may not be obvious from the manual page descriptions.
823
824	<H3><A NAME="caution">Some Notes of Caution</A></H3>
825
826	If you find yourself thinking you need to use <CODE>noraw()</CODE> or
827	<CODE>nocbreak()</CODE>, think again and move carefully. It's probably
828	better design to use <CODE>getstr()</CODE> or one of its relatives to
829	simulate cooked mode. The <CODE>noraw()</CODE> and <CODE>nocbreak()</CODE>
830	functions try to restore cooked mode, but they may end up clobbering
831	some control bits set before you started your application. Also, they
832	have always been poorly documented, and are likely to hurt your
833	application's usability with other curses libraries. <P>
834
835	Bear in mind that <CODE>refresh()</CODE> is a synonym for <CODE>wrefresh(stdscr)</CODE>.
836	Don't try to mix use of <CODE>stdscr</CODE> with use of windows declared
837	by <CODE>newwin()</CODE>; a <CODE>refresh()</CODE> call will blow them off the
838	screen. The right way to handle this is to use <CODE>subwin()</CODE>, or
839	not touch <CODE>stdscr</CODE> at all and tile your screen with declared
840	windows which you then <CODE>wnoutrefresh()</CODE> somewhere in your program
841	event loop, with a single <CODE>doupdate()</CODE> call to trigger actual
842	repainting. <P>
843
844	You are much less likely to run into problems if you design your screen
845	layouts to use tiled rather than overlapping windows. Historically,
846	curses support for overlapping windows has been weak, fragile, and poorly
847	documented. The <CODE>ncurses</CODE> library is not yet an exception to this
848	rule. <P>
849
850	There is a panels library included in the <CODE>ncurses</CODE>
851	distribution that does a pretty good job of strengthening the
852	overlapping-windows facilities. <P>
853
854	Try to avoid using the global variables LINES and COLS. Use
855	<CODE>getmaxyx()</CODE> on the <CODE>stdscr</CODE> context instead. Reason:
856	your code may be ported to run in an environment with window resizes,
857	in which case several screens could be open with different sizes.
858
859	<H3><A NAME="leaving">Temporarily Leaving NCURSES Mode</A></H3>
860
861	Sometimes you will want to write a program that spends most of its time in
862	screen mode, but occasionally returns to ordinary `cooked' mode. A common
863	reason for this is to support shell-out. This behavior is simple to arrange
864	in <CODE>ncurses</CODE>. <P>
865
866	To leave <CODE>ncurses</CODE> mode, call <CODE>endwin()</CODE> as you would if you
867	were intending to terminate the program. This will take the screen back to
868	cooked mode; you can do your shell-out. When you want to return to
869	<CODE>ncurses</CODE> mode, simply call <CODE>refresh()</CODE> or <CODE>doupdate()</CODE>.
870	This will repaint the screen. <P>
871
872	There is a boolean function, <CODE>isendwin()</CODE>, which code can use to
873	test whether <CODE>ncurses</CODE> screen mode is active. It returns <CODE>TRUE</CODE>
874	in the interval between an <CODE>endwin()</CODE> call and the following
875	<CODE>refresh()</CODE>, <CODE>FALSE</CODE> otherwise. <P>
876
877	Here is some sample code for shellout:
878
879	<PRE>
880	addstr("Shelling out...");
881	def_prog_mode(); /* save current tty modes */
882	endwin(); /* restore original tty modes */
883	system("sh"); /* run shell */
884	addstr("returned.\n"); /* prepare return message */
885	refresh(); /* restore save modes, repaint screen */
886	</PRE>
887
888	<H3><A NAME="xterm">Using NCURSES under XTERM</A></H3>
889
890	A resize operation in X sends SIGWINCH to the application running under xterm.
891	The <CODE>ncurses</CODE> library provides an experimental signal
892	handler, but in general does not catch this signal, because it cannot
893	know how you want the screen re-painted. You will usually have to write the
894	SIGWINCH handler yourself. Ncurses can give you some help. <P>
895
896	The easiest way to code your SIGWINCH handler is to have it do an
897	<CODE>endwin</CODE>, followed by an <CODE>refresh</CODE> and a screen repaint you code
898	yourself. The <CODE>refresh</CODE> will pick up the new screen size from the
899	xterm's environment. <P>
900
901	That is the standard way, of course (it even works with some vendor's curses
902	implementations).
903	Its drawback is that it clears the screen to reinitialize the display, and does
904	not resize subwindows which must be shrunk.
905	<CODE>Ncurses</CODE> provides an extension which works better, the
906	<CODE>resizeterm</CODE> function. That function ensures that all windows
907	are limited to the new screen dimensions, and pads <CODE>stdscr</CODE>
908	with blanks if the screen is larger. <P>
909
910	Finally, ncurses can be configured to provide its own SIGWINCH handler,
911	based on <CODE>resizeterm</CODE>.
912
913	<H3><A NAME="screens">Handling Multiple Terminal Screens</A></H3>
914
915	The <CODE>initscr()</CODE> function actually calls a function named
916	<CODE>newterm()</CODE> to do most of its work. If you are writing a program that
917	opens multiple terminals, use <CODE>newterm()</CODE> directly. <P>
918
919	For each call, you will have to specify a terminal type and a pair of file
920	pointers; each call will return a screen reference, and <CODE>stdscr</CODE> will be
921	set to the last one allocated. You will switch between screens with the
922	<CODE>set_term</CODE> call. Note that you will also have to call
923	<CODE>def_shell_mode</CODE> and <CODE>def_prog_mode</CODE> on each tty yourself.
924
925	<H3><A NAME="testing">Testing for Terminal Capabilities</A></H3>
926
927	Sometimes you may want to write programs that test for the presence of various
928	capabilities before deciding whether to go into <CODE>ncurses</CODE> mode. An easy
929	way to do this is to call <CODE>setupterm()</CODE>, then use the functions
930	<CODE>tigetflag()</CODE>, <CODE>tigetnum()</CODE>, and <CODE>tigetstr()</CODE> to do your
931	testing. <P>
932
933	A particularly useful case of this often comes up when you want to
934	test whether a given terminal type should be treated as `smart'
935	(cursor-addressable) or `stupid'. The right way to test this is to see
936	if the return value of <CODE>tigetstr("cup")</CODE> is non-NULL. Alternatively,
937	you can include the <CODE>term.h</CODE> file and test the value of the
938	macro <CODE>cursor_address</CODE>.
939
940	<H3><A NAME="tuning">Tuning for Speed</A></H3>
941
942	Use the <CODE>addchstr()</CODE> family of functions for fast
943	screen-painting of text when you know the text doesn't contain any
944	control characters. Try to make attribute changes infrequent on your
945	screens. Don't use the <CODE>immedok()</CODE> option!
946
947	<H3><A NAME="special">Special Features of NCURSES</A></H3>
948
949	The <CODE>wresize()</CODE> function allows you to resize a window in place.
950	The associated <CODE>resizeterm()</CODE> function simplifies the construction
951	of <a HREF="#xterm">SIGWINCH</a> handlers, for resizing all windows. <P>
952
953	The <CODE>define_key()</CODE> function allows you
954	to define at runtime function-key control sequences which are not in the
955	terminal description.
956	The <CODE>keyok()</CODE> function allows you to temporarily
957	enable or disable interpretation of any function-key control sequence. <P>
958
959	The <CODE>use_default_colors()</CODE> function allows you to construct
960	applications which can use the terminal's default foreground and
961	background colors as an additional "default" color.
962	Several terminal emulators support this feature, which is based on ISO 6429. <P>
963
964	Ncurses supports up 16 colors, unlike SVr4 curses which defines only 8.
965	While most terminals which provide color allow only 8 colors, about
966	a quarter (including XFree86 xterm) support 16 colors.
967
968	<H2><A NAME="compat">Compatibility with Older Versions</A></H2>
969
970	Despite our best efforts, there are some differences between <CODE>ncurses</CODE>
971	and the (undocumented!) behavior of older curses implementations. These arise
972	from ambiguities or omissions in the documentation of the API.
973
974	<H3><A NAME="refbug">Refresh of Overlapping Windows</A></H3>
975
976	If you define two windows A and B that overlap, and then alternately scribble
977	on and refresh them, the changes made to the overlapping region under historic
978	<CODE>curses</CODE> versions were often not documented precisely. <P>
979
980	To understand why this is a problem, remember that screen updates are
981	calculated between two representations of the <EM>entire</EM> display. The
982	documentation says that when you refresh a window, it is first copied to to the
983	virtual screen, and then changes are calculated to update the physical screen
984	(and applied to the terminal). But "copied to" is not very specific, and
985	subtle differences in how copying works can produce different behaviors in the
986	case where two overlapping windows are each being refreshed at unpredictable
987	intervals. <P>
988
989	What happens to the overlapping region depends on what <CODE>wnoutrefresh()</CODE>
990	does with its argument -- what portions of the argument window it copies to the
991	virtual screen. Some implementations do "change copy", copying down only
992	locations in the window that have changed (or been marked changed with
993	<CODE>wtouchln()</CODE> and friends). Some implementations do "entire copy",
994	copying <EM>all</EM> window locations to the virtual screen whether or not
995	they have changed. <P>
996
997	The <CODE>ncurses</CODE> library itself has not always been consistent on this
998	score. Due to a bug, versions 1.8.7 to 1.9.8a did entire copy. Versions
999	1.8.6 and older, and versions 1.9.9 and newer, do change copy. <P>
1000
1001	For most commercial curses implementations, it is not documented and not known
1002	for sure (at least not to the <CODE>ncurses</CODE> maintainers) whether they do
1003	change copy or entire copy. We know that System V release 3 curses has logic
1004	in it that looks like an attempt to do change copy, but the surrounding logic
1005	and data representations are sufficiently complex, and our knowledge
1006	sufficiently indirect, that it's hard to know whether this is reliable.
1007
1008	It is not clear what the SVr4 documentation and XSI standard intend. The XSI
1009	Curses standard barely mentions wnoutrefresh(); the SVr4 documents seem to be
1010	describing entire-copy, but it is possible with some effort and straining to
1011	read them the other way. <P>
1012
1013	It might therefore be unwise to rely on either behavior in programs that might
1014	have to be linked with other curses implementations. Instead, you can do an
1015	explicit <CODE>touchwin()</CODE> before the <CODE>wnoutrefresh()</CODE> call to
1016	guarantee an entire-contents copy anywhere. <P>
1017
1018	The really clean way to handle this is to use the panels library. If,
1019	when you want a screen update, you do <CODE>update_panels()</CODE>, it will
1020	do all the necessary <CODE>wnoutrefresh()</CODE> calls for whatever panel
1021	stacking order you have defined. Then you can do one <CODE>doupdate()</CODE>
1022	and there will be a <EM>single</EM> burst of physical I/O that will do
1023	all your updates.
1024
1025	<H3><A NAME="backbug">Background Erase</A></H3>
1026
1027	If you have been using a very old versions of <CODE>ncurses</CODE> (1.8.7 or
1028	older) you may be surprised by the behavior of the erase functions. In older
1029	versions, erased areas of a window were filled with a blank modified by the
1030	window's current attribute (as set by <STRONG>wattrset()</STRONG>, <STRONG>wattron()</STRONG>,
1031	<STRONG>wattroff()</STRONG> and friends). <P>
1032
1033	In newer versions, this is not so. Instead, the attribute of erased blanks
1034	is normal unless and until it is modified by the functions <CODE>bkgdset()</CODE>
1035	or <CODE>wbkgdset()</CODE>. <P>
1036
1037	This change in behavior conforms <CODE>ncurses</CODE> to System V Release 4 and
1038	the XSI Curses standard.
1039
1040	<H2><A NAME="xsifuncs">XSI Curses Conformance</A></H2>
1041
1042	The <CODE>ncurses</CODE> library is intended to be base-level conformant with the
1043	XSI Curses standard from X/Open. Many extended-level features (in fact, almost
1044	all features not directly concerned with wide characters and
1045	internationalization) are also supported. <P>
1046
1047	One effect of XSI conformance is the change in behavior described under
1048	<A HREF="#backbug">"Background Erase -- Compatibility with Old Versions"</A>. <P>
1049
1050	Also, <CODE>ncurses</CODE> meets the XSI requirement that every macro
1051	entry point have a corresponding function which may be linked (and
1052	will be prototype-checked) if the macro definition is disabled with
1053	<CODE>#undef</CODE>.
1054
1055	<H1><A NAME="panels">The Panels Library</A></H1>
1056
1057	The <CODE>ncurses</CODE> library by itself provides good support for screen
1058	displays in which the windows are tiled (non-overlapping). In the more
1059	general case that windows may overlap, you have to use a series of
1060	<CODE>wnoutrefresh()</CODE> calls followed by a <CODE>doupdate()</CODE>, and be
1061	careful about the order you do the window refreshes in. It has to be
1062	bottom-upwards, otherwise parts of windows that should be obscured will
1063	show through. <P>
1064
1065	When your interface design is such that windows may dive deeper into the
1066	visibility stack or pop to the top at runtime, the resulting book-keeping
1067	can be tedious and difficult to get right. Hence the panels library. <P>
1068
1069	The <CODE>panel</CODE> library first appeared in AT&T System V. The
1070	version documented here is the <CODE>panel</CODE> code distributed
1071	with <CODE>ncurses</CODE>.
1072
1073	<H2><A NAME="pcompile">Compiling With the Panels Library</A></H2>
1074
1075	Your panels-using modules must import the panels library declarations with
1076
1077	<PRE>
1078	#include <panel.h>
1079	</PRE>
1080
1081	and must be linked explicitly with the panels library using an
1082	<CODE>-lpanel</CODE> argument. Note that they must also link the
1083	<CODE>ncurses</CODE> library with <CODE>-lncurses</CODE>. Many linkers
1084	are two-pass and will accept either order, but it is still good practice
1085	to put <CODE>-lpanel</CODE> first and <CODE>-lncurses</CODE> second.
1086
1087	<H2><A NAME="poverview">Overview of Panels</A></H2>
1088
1089	A panel object is a window that is implicitly treated as part of a
1090	<DFN>deck</DFN> including all other panel objects. The deck has an implicit
1091	bottom-to-top visibility order. The panels library includes an update
1092	function (analogous to <CODE>refresh()</CODE>) that displays all panels in the
1093	deck in the proper order to resolve overlaps. The standard window,
1094	<CODE>stdscr</CODE>, is considered below all panels. <P>
1095
1096	Details on the panels functions are available in the man pages. We'll just
1097	hit the highlights here. <P>
1098
1099	You create a panel from a window by calling <CODE>new_panel()</CODE> on a
1100	window pointer. It then becomes the top of the deck. The panel's window
1101	is available as the value of <CODE>panel_window()</CODE> called with the
1102	panel pointer as argument.<P>
1103
1104	You can delete a panel (removing it from the deck) with <CODE>del_panel</CODE>.
1105	This will not deallocate the associated window; you have to do that yourself.
1106
1107	You can replace a panel's window with a different window by calling
1108	<CODE>replace_window</CODE>. The new window may be of different size;
1109	the panel code will re-compute all overlaps. This operation doesn't
1110	change the panel's position in the deck. <P>
1111
1112	To move a panel's window, use <CODE>move_panel()</CODE>. The
1113	<CODE>mvwin()</CODE> function on the panel's window isn't sufficient because it
1114	doesn't update the panels library's representation of where the windows are.
1115	This operation leaves the panel's depth, contents, and size unchanged. <P>
1116
1117	Two functions (<CODE>top_panel()</CODE>, <CODE>bottom_panel()</CODE>) are
1118	provided for rearranging the deck. The first pops its argument window to the
1119	top of the deck; the second sends it to the bottom. Either operation leaves
1120	the panel's screen location, contents, and size unchanged. <P>
1121
1122	The function <CODE>update_panels()</CODE> does all the
1123	<CODE>wnoutrefresh()</CODE> calls needed to prepare for
1124	<CODE>doupdate()</CODE> (which you must call yourself, afterwards). <P>
1125
1126	Typically, you will want to call <CODE>update_panels()</CODE> and
1127	<CODE>doupdate()</CODE> just before accepting command input, once in each cycle
1128	of interaction with the user. If you call <CODE>update_panels()</CODE> after
1129	each and every panel write, you'll generate a lot of unnecessary refresh
1130	activity and screen flicker.
1131
1132	<H2><A NAME="pstdscr">Panels, Input, and the Standard Screen</A></H2>
1133
1134	You shouldn't mix <CODE>wnoutrefresh()</CODE> or <CODE>wrefresh()</CODE>
1135	operations with panels code; this will work only if the argument window
1136	is either in the top panel or unobscured by any other panels. <P>
1137
1138	The <CODE>stsdcr</CODE> window is a special case. It is considered below all
1139	panels. Because changes to panels may obscure parts of <CODE>stdscr</CODE>,
1140	though, you should call <CODE>update_panels()</CODE> before
1141	<CODE>doupdate()</CODE> even when you only change <CODE>stdscr</CODE>. <P>
1142
1143	Note that <CODE>wgetch</CODE> automatically calls <CODE>wrefresh</CODE>.
1144	Therefore, before requesting input from a panel window, you need to be sure
1145	that the panel is totally unobscured. <P>
1146
1147	There is presently no way to display changes to one obscured panel without
1148	repainting all panels.
1149
1150	<H2><A NAME="hiding">Hiding Panels</A></H2>
1151
1152	It's possible to remove a panel from the deck temporarily; use
1153	<CODE>hide_panel</CODE> for this. Use <CODE>show_panel()</CODE> to render it
1154	visible again. The predicate function <CODE>panel_hidden</CODE>
1155	tests whether or not a panel is hidden. <P>
1156
1157	The <CODE>panel_update</CODE> code ignores hidden panels. You cannot do
1158	<CODE>top_panel()</CODE> or <CODE>bottom_panel</CODE> on a hidden panel().
1159	Other panels operations are applicable.
1160
1161	<H2><A NAME="pmisc">Miscellaneous Other Facilities</A></H2>
1162
1163	It's possible to navigate the deck using the functions
1164	<CODE>panel_above()</CODE> and <CODE>panel_below</CODE>. Handed a panel
1165	pointer, they return the panel above or below that panel. Handed
1166	<CODE>NULL</CODE>, they return the bottom-most or top-most panel. <P>
1167
1168	Every panel has an associated user pointer, not used by the panel code, to
1169	which you can attach application data. See the man page documentation
1170	of <CODE>set_panel_userptr()</CODE> and <CODE>panel_userptr</CODE> for
1171	details.
1172
1173	<H1><A NAME="menu">The Menu Library</A></H1>
1174
1175	A menu is a screen display that assists the user to choose some subset
1176	of a given set of items. The <CODE>menu</CODE> library is a curses
1177	extension that supports easy programming of menu hierarchies with a
1178	uniform but flexible interface. <P>
1179
1180	The <CODE>menu</CODE> library first appeared in AT&T System V. The
1181	version documented here is the <CODE>menu</CODE> code distributed
1182	with <CODE>ncurses</CODE>.
1183
1184	<H2><A NAME="mcompile">Compiling With the menu Library</A></H2>
1185
1186	Your menu-using modules must import the menu library declarations with
1187
1188	<PRE>
1189	#include <menu.h>
1190	</PRE>
1191
1192	and must be linked explicitly with the menus library using an
1193	<CODE>-lmenu</CODE> argument. Note that they must also link the
1194	<CODE>ncurses</CODE> library with <CODE>-lncurses</CODE>. Many linkers
1195	are two-pass and will accept either order, but it is still good practice
1196	to put <CODE>-lmenu</CODE> first and <CODE>-lncurses</CODE> second.
1197
1198	<H2><A NAME="moverview">Overview of Menus</A></H2>
1199
1200	The menus created by this library consist of collections of
1201	<DFN>items</DFN> including a name string part and a description string
1202	part. To make menus, you create groups of these items and connect
1203	them with menu frame objects. <P>
1204
1205	The menu can then by <DFN>posted</DFN>, that is written to an
1206	associated window. Actually, each menu has two associated windows; a
1207	containing window in which the programmer can scribble titles or
1208	borders, and a subwindow in which the menu items proper are displayed.
1209	If this subwindow is too small to display all the items, it will be a
1210	scrollable viewport on the collection of items. <P>
1211
1212	A menu may also be <DFN>unposted</DFN> (that is, undisplayed), and finally
1213	freed to make the storage associated with it and its items available for
1214	re-use. <P>
1215
1216	The general flow of control of a menu program looks like this:
1217
1218	<OL>
1219	<LI>Initialize <CODE>curses</CODE>.
1220	<LI>Create the menu items, using <CODE>new_item()</CODE>.
1221	<LI>Create the menu using <CODE>new_menu()</CODE>.
1222	<LI>Post the menu using <CODE>post_menu()</CODE>.
1223	<LI>Refresh the screen.
1224	<LI>Process user requests via an input loop.
1225	<LI>Unpost the menu using <CODE>unpost_menu()</CODE>.
1226	<LI>Free the menu, using <CODE>free_menu()</CODE>.
1227	<LI>Free the items using <CODE>free_item()</CODE>.
1228	<LI>Terminate <CODE>curses</CODE>.
1229	</OL>
1230
1231	<H2><A NAME="mselect">Selecting items</A></H2>
1232
1233	Menus may be multi-valued or (the default) single-valued (see the manual
1234	page <CODE>menu_opts(3x)</CODE> to see how to change the default).
1235	Both types always have a <DFN>current item</DFN>. <P>
1236
1237	From a single-valued menu you can read the selected value simply by looking
1238	at the current item. From a multi-valued menu, you get the selected set
1239	by looping through the items applying the <CODE>item_value()</CODE>
1240	predicate function. Your menu-processing code can use the function
1241	<CODE>set_item_value()</CODE> to flag the items in the select set. <P>
1242
1243	Menu items can be made unselectable using <CODE>set_item_opts()</CODE>
1244	or <CODE>item_opts_off()</CODE> with the <CODE>O_SELECTABLE</CODE>
1245	argument. This is the only option so far defined for menus, but it
1246	is good practice to code as though other option bits might be on.
1247
1248	<H2><A NAME="mdisplay">Menu Display</A></H2>
1249
1250	The menu library calculates a minimum display size for your window, based
1251	on the following variables:
1252
1253	<UL>
1254	<LI>The number and maximum length of the menu items
1255	<LI>Whether the O_ROWMAJOR option is enabled
1256	<LI>Whether display of descriptions is enabled
1257	<LI>Whatever menu format may have been set by the programmer
1258	<LI>The length of the menu mark string used for highlighting selected items
1259	</UL>
1260
1261	The function <CODE>set_menu_format()</CODE> allows you to set the
1262	maximum size of the viewport or <DFN>menu page</DFN> that will be used
1263	to display menu items. You can retrieve any format associated with a
1264	menu with <CODE>menu_format()</CODE>. The default format is rows=16,
1265	columns=1. <P>
1266
1267	The actual menu page may be smaller than the format size. This depends
1268	on the item number and size and whether O_ROWMAJOR is on. This option
1269	(on by default) causes menu items to be displayed in a `raster-scan'
1270	pattern, so that if more than one item will fit horizontally the first
1271	couple of items are side-by-side in the top row. The alternative is
1272	column-major display, which tries to put the first several items in
1273	the first column. <P>
1274
1275	As mentioned above, a menu format not large enough to allow all items to fit
1276	on-screen will result in a menu display that is vertically scrollable. <P>
1277	You can scroll it with requests to the menu driver, which will be described
1278	in the section on <A HREF="#minput">menu input handling</A>. <P>
1279
1280	Each menu has a <DFN>mark string</DFN> used to visually tag selected items;
1281	see the <CODE>menu_mark(3x)</CODE> manual page for details. The mark
1282	string length also influences the menu page size. <P>
1283
1284	The function <CODE>scale_menu()</CODE> returns the minimum display size
1285	that the menu code computes from all these factors.
1286
1287	There are other menu display attributes including a select attribute,
1288	an attribute for selectable items, an attribute for unselectable items,
1289	and a pad character used to separate item name text from description
1290	text. These have reasonable defaults which the library allows you to
1291	change (see the <CODE>menu_attribs(3x)</CODE> manual page.
1292
1293	<H2><A NAME="mwindows">Menu Windows</A></H2>
1294
1295	Each menu has, as mentioned previously, a pair of associated windows.
1296	Both these windows are painted when the menu is posted and erased when
1297	the menu is unposted. <P>
1298
1299	The outer or frame window is not otherwise touched by the menu
1300	routines. It exists so the programmer can associate a title, a
1301	border, or perhaps help text with the menu and have it properly
1302	refreshed or erased at post/unpost time. The inner window or
1303	<DFN>subwindow</DFN> is where the current menu page is displayed. <P>
1304
1305	By default, both windows are <CODE>stdscr</CODE>. You can set them with the
1306	functions in <CODE>menu_win(3x)</CODE>. <P>
1307
1308	When you call <CODE>post_menu()</CODE>, you write the menu to its
1309	subwindow. When you call <CODE>unpost_menu()</CODE>, you erase the
1310	subwindow, However, neither of these actually modifies the screen. To
1311	do that, call <CODE>wrefresh()</CODE> or some equivalent.
1312
1313	<H2><A NAME="minput">Processing Menu Input</A></H2>
1314
1315	The main loop of your menu-processing code should call
1316	<CODE>menu_driver()</CODE> repeatedly. The first argument of this routine
1317	is a menu pointer; the second is a menu command code. You should write an
1318	input-fetching routine that maps input characters to menu command codes, and
1319	pass its output to <CODE>menu_driver()</CODE>. The menu command codes are
1320	fully documented in <CODE>menu_driver(3x)</CODE>. <P>
1321
1322	The simplest group of command codes is <CODE>REQ_NEXT_ITEM</CODE>,
1323	<CODE>REQ_PREV_ITEM</CODE>, <CODE>REQ_FIRST_ITEM</CODE>,
1324	<CODE>REQ_LAST_ITEM</CODE>, <CODE>REQ_UP_ITEM</CODE>,
1325	<CODE>REQ_DOWN_ITEM</CODE>, <CODE>REQ_LEFT_ITEM</CODE>,
1326	<CODE>REQ_RIGHT_ITEM</CODE>. These change the currently selected
1327	item. These requests may cause scrolling of the menu page if it only
1328	partially displayed. <P>
1329
1330	There are explicit requests for scrolling which also change the
1331	current item (because the select location does not change, but the
1332	item there does). These are <CODE>REQ_SCR_DLINE</CODE>,
1333	<CODE>REQ_SCR_ULINE</CODE>, <CODE>REQ_SCR_DPAGE</CODE>, and
1334	<CODE>REQ_SCR_UPAGE</CODE>. <P>
1335
1336	The <CODE>REQ_TOGGLE_ITEM</CODE> selects or deselects the current item.
1337	It is for use in multi-valued menus; if you use it with <CODE>O_ONEVALUE</CODE>
1338	on, you'll get an error return (<CODE>E_REQUEST_DENIED</CODE>). <P>
1339
1340	Each menu has an associated pattern buffer. The
1341	<CODE>menu_driver()</CODE> logic tries to accumulate printable ASCII
1342	characters passed in in that buffer; when it matches a prefix of an
1343	item name, that item (or the next matching item) is selected. If
1344	appending a character yields no new match, that character is deleted
1345	from the pattern buffer, and <CODE>menu_driver()</CODE> returns
1346	<CODE>E_NO_MATCH</CODE>. <P>
1347
1348	Some requests change the pattern buffer directly:
1349	<CODE>REQ_CLEAR_PATTERN</CODE>, <CODE>REQ_BACK_PATTERN</CODE>,
1350	<CODE>REQ_NEXT_MATCH</CODE>, <CODE>REQ_PREV_MATCH</CODE>. The latter
1351	two are useful when pattern buffer input matches more than one item
1352	in a multi-valued menu. <P>
1353
1354	Each successful scroll or item navigation request clears the pattern
1355	buffer. It is also possible to set the pattern buffer explicitly
1356	with <CODE>set_menu_pattern()</CODE>. <P>
1357
1358	Finally, menu driver requests above the constant <CODE>MAX_COMMAND</CODE>
1359	are considered application-specific commands. The <CODE>menu_driver()</CODE>
1360	code ignores them and returns <CODE>E_UNKNOWN_COMMAND</CODE>.
1361
1362	<H2><A NAME="mmisc">Miscellaneous Other Features</A></H2>
1363
1364	Various menu options can affect the processing and visual appearance
1365	and input processing of menus. See <CODE>menu_opts(3x) for
1366	details.</CODE> <P>
1367
1368	It is possible to change the current item from application code; this
1369	is useful if you want to write your own navigation requests. It is
1370	also possible to explicitly set the top row of the menu display. See
1371	<CODE>mitem_current(3x)</CODE>.
1372
1373	If your application needs to change the menu subwindow cursor for
1374	any reason, <CODE>pos_menu_cursor()</CODE> will restore it to the
1375	correct location for continuing menu driver processing. <P>
1376
1377	It is possible to set hooks to be called at menu initialization and
1378	wrapup time, and whenever the selected item changes. See
1379	<CODE>menu_hook(3x)</CODE>. <P>
1380
1381	Each item, and each menu, has an associated user pointer on which you
1382	can hang application data. See <CODE>mitem_userptr(3x)</CODE> and
1383	<CODE>menu_userptr(3x)</CODE>.
1384
1385	<H1><A NAME="form">The Forms Library</A></H1>
1386
1387	The <CODE>form</CODE> library is a curses extension that supports easy
1388	programming of on-screen forms for data entry and program control. <P>
1389
1390	The <CODE>form</CODE> library first appeared in AT&T System V. The
1391	version documented here is the <CODE>form</CODE> code distributed
1392	with <CODE>ncurses</CODE>.
1393
1394	<H2><A NAME="fcompile">Compiling With the form Library</A></H2>
1395
1396	Your form-using modules must import the form library declarations with
1397
1398	<PRE>
1399	#include <form.h>
1400	</PRE>
1401
1402	and must be linked explicitly with the forms library using an
1403	<CODE>-lform</CODE> argument. Note that they must also link the
1404	<CODE>ncurses</CODE> library with <CODE>-lncurses</CODE>. Many linkers
1405	are two-pass and will accept either order, but it is still good practice
1406	to put <CODE>-lform</CODE> first and <CODE>-lncurses</CODE> second.
1407
1408	<H2><A NAME="foverview">Overview of Forms</A></H2>
1409
1410	A form is a collection of fields; each field may be either a label
1411	(explanatory text) or a data-entry location. Long forms may be
1412	segmented into pages; each entry to a new page clears the screen. <P>
1413	To make forms, you create groups of fields and connect them with form
1414	frame objects; the form library makes this relatively simple. <P>
1415
1416	Once defined, a form can be <DFN>posted</DFN>, that is written to an
1417	associated window. Actually, each form has two associated windows; a
1418	containing window in which the programmer can scribble titles or
1419	borders, and a subwindow in which the form fields proper are displayed. <P>
1420
1421	As the form user fills out the posted form, navigation and editing
1422	keys support movement between fields, editing keys support modifying
1423	field, and plain text adds to or changes data in a current field. The
1424	form library allows you (the forms designer) to bind each navigation
1425	and editing key to any keystroke accepted by <CODE>curses</CODE>
1426
1427	Fields may have validation conditions on them, so that they check input
1428	data for type and value. The form library supplies a rich set of
1429	pre-defined field types, and makes it relatively easy to define new ones. <P>
1430
1431	Once its transaction is completed (or aborted), a form may be
1432	<DFN>unposted</DFN> (that is, undisplayed), and finally freed to make
1433	the storage associated with it and its items available for re-use. <P>
1434
1435	The general flow of control of a form program looks like this:
1436
1437	<OL>
1438	<LI>Initialize <CODE>curses</CODE>.
1439	<LI>Create the form fields, using <CODE>new_field()</CODE>.
1440	<LI>Create the form using <CODE>new_form()</CODE>.
1441	<LI>Post the form using <CODE>post_form()</CODE>.
1442	<LI>Refresh the screen.
1443	<LI>Process user requests via an input loop.
1444	<LI>Unpost the form using <CODE>unpost_form()</CODE>.
1445	<LI>Free the form, using <CODE>free_form()</CODE>.
1446	<LI>Free the fields using <CODE>free_field()</CODE>.
1447	<LI>Terminate <CODE>curses</CODE>.
1448	</OL>
1449
1450	Note that this looks much like a menu program; the form library handles
1451	tasks which are in many ways similar, and its interface was obviously
1452	designed to resemble that of the <A HREF="#menu">menu library</A>
1453	wherever possible. <P>
1454
1455	In forms programs, however, the `process user requests' is somewhat more
1456	complicated than for menus. Besides menu-like navigation operations,
1457	the menu driver loop has to support field editing and data validation.
1458
1459	<H2><A NAME="fcreate">Creating and Freeing Fields and Forms</A></H2>
1460
1461	The basic function for creating fields is <CODE>new_field()</CODE>:
1462
1463	<PRE>
1464	FIELD new_field(int height, int width, / new field size */
1465	int top, int left, /* upper left corner */
1466	int offscreen, /* number of offscreen rows */
1467	int nbuf); /* number of working buffers */
1468	</PRE>
1469
1470	Menu items always occupy a single row, but forms fields may have
1471	multiple rows. So <CODE>new_field()</CODE> requires you to specify a
1472	width and height (the first two arguments, which mist both be greater
1473	than zero). <P>
1474
1475	You must also specify the location of the field's upper left corner on
1476	the screen (the third and fourth arguments, which must be zero or
1477	greater). Note that these coordinates are relative to the form
1478	subwindow, which will coincide with <CODE>stdscr</CODE> by default but
1479	need not be <CODE>stdscr</CODE> if you've done an explicit
1480	<CODE>set_form_win()</CODE> call. <P>
1481
1482	The fifth argument allows you to specify a number of off-screen rows. If
1483	this is zero, the entire field will always be displayed. If it is
1484	nonzero, the form will be scrollable, with only one screen-full (initially
1485	the top part) displayed at any given time. If you make a field dynamic
1486	and grow it so it will no longer fit on the screen, the form will become
1487	scrollable even if the <CODE>offscreen</CODE> argument was initially zero. <P>
1488
1489	The forms library allocates one working buffer per field; the size of
1490	each buffer is <CODE>((height + offscreen)*width + 1</CODE>, one character
1491	for each position in the field plus a NUL terminator. The sixth
1492	argument is the number of additional data buffers to allocate for the
1493	field; your application can use them for its own purposes.
1494
1495	<PRE>
1496	FIELD dup_field(FIELD field, /* field to copy */
1497	int top, int left); /* location of new copy */
1498	</PRE>
1499
1500	The function <CODE>dup_field()</CODE> duplicates an existing field at a
1501	new location. Size and buffering information are copied; some
1502	attribute flags and status bits are not (see the
1503	<CODE>form_field_new(3X)</CODE> for details).
1504
1505	<PRE>
1506	FIELD link_field(FIELD field, /* field to copy */
1507	int top, int left); /* location of new copy */
1508	</PRE>
1509
1510	The function <CODE>link_field()</CODE> also duplicates an existing field
1511	at a new location. The difference from <CODE>dup_field()</CODE> is that
1512	it arranges for the new field's buffer to be shared with the old one. <P>
1513
1514	Besides the obvious use in making a field editable from two different
1515	form pages, linked fields give you a way to hack in dynamic labels. If
1516	you declare several fields linked to an original, and then make them
1517	inactive, changes from the original will still be propagated to the
1518	linked fields. <P>
1519
1520	As with duplicated fields, linked fields have attribute bits separate
1521	from the original. <P>
1522
1523	As you might guess, all these field-allocations return <CODE>NULL</CODE> if
1524	the field allocation is not possible due to an out-of-memory error or
1525	out-of-bounds arguments. <P>
1526
1527	To connect fields to a form, use
1528
1529	<PRE>
1530	FORM new_form(FIELD *fields);
1531	</PRE>
1532
1533	This function expects to see a NULL-terminated array of field pointers.
1534	Said fields are connected to a newly-allocated form object; its address
1535	is returned (or else NULL if the allocation fails). <P>
1536
1537	Note that <CODE>new_field()</CODE> does <EM>not</EM> copy the pointer array
1538	into private storage; if you modify the contents of the pointer array
1539	during forms processing, all manner of bizarre things might happen. Also
1540	note that any given field may only be connected to one form. <P>
1541
1542	The functions <CODE>free_field()</CODE> and <CODE>free_form</CODE> are available
1543	to free field and form objects. It is an error to attempt to free a field
1544	connected to a form, but not vice-versa; thus, you will generally free
1545	your form objects first.
1546
1547	<H2><A NAME="fattributes">Fetching and Changing Field Attributes</A></H2>
1548
1549	Each form field has a number of location and size attributes
1550	associated with it. There are other field attributes used to control
1551	display and editing of the field. Some (for example, the <CODE>O_STATIC</CODE> bit)
1552	involve sufficient complications to be covered in sections of their own
1553	later on. We cover the functions used to get and set several basic
1554	attributes here. <P>
1555
1556	When a field is created, the attributes not specified by the
1557	<CODE>new_field</CODE> function are copied from an invisible system
1558	default field. In attribute-setting and -fetching functions, the
1559	argument NULL is taken to mean this field. Changes to it persist
1560	as defaults until your forms application terminates.
1561
1562	<H3><A NAME="fsizes">Fetching Size and Location Data</A></H3>
1563
1564	You can retrieve field sizes and locations through:
1565
1566	<PRE>
1567	int field_info(FIELD field, / field from which to fetch */
1568	int height, int width, /* field size */
1569	int top, int left, /* upper left corner */
1570	int offscreen, / number of offscreen rows */
1571	int nbuf); / number of working buffers */
1572	</PRE>
1573
1574	This function is a sort of inverse of <CODE>new_field()</CODE>; instead of
1575	setting size and location attributes of a new field, it fetches them
1576	from an existing one.
1577
1578	<H3><A NAME="flocation">Changing the Field Location</A></H3>
1579
1580	It is possible to move a field's location on the screen:
1581
1582	<PRE>
1583	int move_field(FIELD field, / field to alter */
1584	int top, int left); /* new upper-left corner */
1585	</PRE>
1586
1587	You can, of course. query the current location through <CODE>field_info()</CODE>.
1588
1589	<H3><A NAME="fjust">The Justification Attribute</A></H3>
1590
1591	One-line fields may be unjustified, justified right, justified left,
1592	or centered. Here is how you manipulate this attribute:
1593
1594	<PRE>
1595	int set_field_just(FIELD field, / field to alter */
1596	int justmode); /* mode to set */
1597
1598	int field_just(FIELD field); / fetch mode of field */
1599	</PRE>
1600
1601	The mode values accepted and returned by this functions are
1602	preprocessor macros <CODE>NO_JUSTIFICATION</CODE>, <CODE>JUSTIFY_RIGHT</CODE>,
1603	<CODE>JUSTIFY_LEFT</CODE>, or <CODE>JUSTIFY_CENTER</CODE>.
1604
1605	<H3><A NAME="fdispatts">Field Display Attributes</A></H3>
1606
1607	For each field, you can set a foreground attribute for entered
1608	characters, a background attribute for the entire field, and a pad
1609	character for the unfilled portion of the field. You can also
1610	control pagination of the form. <P>
1611
1612	This group of four field attributes controls the visual appearance
1613	of the field on the screen, without affecting in any way the data
1614	in the field buffer.
1615
1616	<PRE>
1617	int set_field_fore(FIELD field, / field to alter */
1618	chtype attr); /* attribute to set */
1619
1620	chtype field_fore(FIELD field); / field to query */
1621
1622	int set_field_back(FIELD field, / field to alter */
1623	chtype attr); /* attribute to set */
1624
1625	chtype field_back(FIELD field); / field to query */
1626
1627	int set_field_pad(FIELD field, / field to alter */
1628	int pad); /* pad character to set */
1629
1630	chtype field_pad(FIELD *field);
1631
1632	int set_new_page(FIELD field, / field to alter */
1633	int flag); /* TRUE to force new page */
1634
1635	chtype new_page(FIELD field); / field to query */
1636	</PRE>
1637
1638	The attributes set and returned by the first four functions are normal
1639	<CODE>curses(3x)</CODE> display attribute values (<CODE>A_STANDOUT</CODE>,
1640	<CODE>A_BOLD</CODE>, <CODE>A_REVERSE</CODE> etc).
1641
1642	The page bit of a field controls whether it is displayed at the start of
1643	a new form screen.
1644
1645	<H3><A NAME="foptions">Field Option Bits</A></H3>
1646
1647	There is also a large collection of field option bits you can set to control
1648	various aspects of forms processing. You can manipulate them with these
1649	functions:
1650
1651	<PRE>
1652	int set_field_opts(FIELD field, / field to alter */
1653	int attr); /* attribute to set */
1654
1655	int field_opts_on(FIELD field, / field to alter */
1656	int attr); /* attributes to turn on */
1657
1658	int field_opts_off(FIELD field, / field to alter */
1659	int attr); /* attributes to turn off */
1660
1661	int field_opts(FIELD field); / field to query */
1662	</PRE>
1663
1664	By default, all options are on. Here are the available option bits:
1665	<DL>
1666	<DT> O_VISIBLE
1667	<DD> Controls whether the field is visible on the screen. Can be used
1668	during form processing to hide or pop up fields depending on the value
1669	of parent fields.
1670	<DT> O_ACTIVE
1671	<DD> Controls whether the field is active during forms processing (i.e.
1672	visited by form navigation keys). Can be used to make labels or derived
1673	fields with buffer values alterable by the forms application, not the user.
1674	<DT> O_PUBLIC
1675	<DD> Controls whether data is displayed during field entry. If this option is
1676	turned off on a field, the library will accept and edit data in that field,
1677	but it will not be displayed and the visible field cursor will not move.
1678	You can turn off the O_PUBLIC bit to define password fields.
1679	<DT> O_EDIT
1680	<DD> Controls whether the field's data can be modified. When this option is
1681	off, all editing requests except <CODE>REQ_PREV_CHOICE</CODE> and
1682	<CODE>REQ_NEXT_CHOICE</CODE> will fail. Such read-only fields may be useful for
1683	help messages.
1684	<DT> O_WRAP
1685	<DD> Controls word-wrapping in multi-line fields. Normally, when any
1686	character of a (blank-separated) word reaches the end of the current line, the
1687	entire word is wrapped to the next line (assuming there is one). When this
1688	option is off, the word will be split across the line break.
1689	<DT> O_BLANK
1690	<DD> Controls field blanking. When this option is on, entering a character at
1691	the first field position erases the entire field (except for the just-entered
1692	character).
1693	<DT> O_AUTOSKIP
1694	<DD> Controls automatic skip to next field when this one fills. Normally,
1695	when the forms user tries to type more data into a field than will fit,
1696	the editing location jumps to next field. When this option is off, the
1697	user's cursor will hang at the end of the field. This option is ignored
1698	in dynamic fields that have not reached their size limit.
1699	<DT> O_NULLOK
1700	<DD> Controls whether <A HREF="#fvalidation">validation</A> is applied to
1701	blank fields. Normally, it is not; the user can leave a field blank
1702	without invoking the usual validation check on exit. If this option is
1703	off on a field, exit from it will invoke a validation check.
1704	<DT> O_PASSOK
1705	<DD> Controls whether validation occurs on every exit, or only after
1706	the field is modified. Normally the latter is true. Setting O_PASSOK
1707	may be useful if your field's validation function may change during
1708	forms processing.
1709	<DT> O_STATIC
1710	<DD> Controls whether the field is fixed to its initial dimensions. If you
1711	turn this off, the field becomes <A HREF="#fdynamic">dynamic</A> and will
1712	stretch to fit entered data.
1713	</DL>
1714
1715	A field's options cannot be changed while the field is currently selected.
1716	However, options may be changed on posted fields that are not current. <P>
1717
1718	The option values are bit-masks and can be composed with logical-or in
1719	the obvious way.
1720
1721	<H2><A NAME="fstatus">Field Status</A></H2>
1722
1723	Every field has a status flag, which is set to FALSE when the field is
1724	created and TRUE when the value in field buffer 0 changes. This flag can
1725	be queried and set directly:
1726
1727	<PRE>
1728	int set_field_status(FIELD field, / field to alter */
1729	int status); /* mode to set */
1730
1731	int field_status(FIELD field); / fetch mode of field */
1732	</PRE>
1733
1734	Setting this flag under program control can be useful if you use the same
1735	form repeatedly, looking for modified fields each time. <P>
1736
1737	Calling <CODE>field_status()</CODE> on a field not currently selected
1738	for input will return a correct value. Calling <CODE>field_status()</CODE> on a
1739	field that is currently selected for input may not necessarily give a
1740	correct field status value, because entered data isn't necessarily copied to
1741	buffer zero before the exit validation check.
1742
1743	To guarantee that the returned status value reflects reality, call
1744	<CODE>field_status()</CODE> either (1) in the field's exit validation check
1745	routine, (2) from the field's or form's initialization or termination
1746	hooks, or (3) just after a <CODE>REQ_VALIDATION</CODE> request has been
1747	processed by the forms driver.
1748
1749	<H2><A NAME="fuser">Field User Pointer</A></H2>
1750
1751	Each field structure contains one character pointer slot that is not used
1752	by the forms library. It is intended to be used by applications to store
1753	private per-field data. You can manipulate it with:
1754
1755	<PRE>
1756	int set_field_userptr(FIELD field, / field to alter */
1757	char userptr); / mode to set */
1758
1759	char field_userptr(FIELD field); /* fetch mode of field */
1760	</PRE>
1761
1762	(Properly, this user pointer field ought to have <CODE>(void *)</CODE> type.
1763	The <CODE>(char *)</CODE> type is retained for System V compatibility.) <P>
1764
1765	It is valid to set the user pointer of the default field (with a
1766	<CODE>set_field_userptr()</CODE> call passed a NULL field pointer.)
1767	When a new field is created, the default-field user pointer is copied
1768	to initialize the new field's user pointer.
1769
1770	<H2><A NAME="fdynamic">Variable-Sized Fields</A></H2>
1771
1772	Normally, a field is fixed at the size specified for it at creation
1773	time. If, however, you turn off its O_STATIC bit, it becomes
1774	<DFN>dynamic</DFN> and will automatically resize itself to accommodate
1775	data as it is entered. If the field has extra buffers associated with it,
1776	they will grow right along with the main input buffer. <P>
1777
1778	A one-line dynamic field will have a fixed height (1) but variable
1779	width, scrolling horizontally to display data within the field area as
1780	originally dimensioned and located. A multi-line dynamic field will
1781	have a fixed width, but variable height (number of rows), scrolling
1782	vertically to display data within the field area as originally
1783	dimensioned and located. <P>
1784
1785	Normally, a dynamic field is allowed to grow without limit. But it is
1786	possible to set an upper limit on the size of a dynamic field. You do
1787	it with this function:
1788
1789	<PRE>
1790	int set_max_field(FIELD field, / field to alter (may not be NULL) */
1791	int max_size); /* upper limit on field size */
1792	</PRE>
1793
1794	If the field is one-line, <CODE>max_size</CODE> is taken to be a column size
1795	limit; if it is multi-line, it is taken to be a line size limit. To disable
1796	any limit, use an argument of zero. The growth limit can be changed whether
1797	or not the O_STATIC bit is on, but has no effect until it is. <P>
1798
1799	The following properties of a field change when it becomes dynamic:
1800
1801	<UL>
1802	<LI>If there is no growth limit, there is no final position of the field;
1803	therefore <CODE>O_AUTOSKIP</CODE> and <CODE>O_NL_OVERLOAD</CODE> are ignored.
1804	<LI>Field justification will be ignored (though whatever justification is
1805	set up will be retained internally and can be queried).
1806	<LI>The <CODE>dup_field()</CODE> and <CODE>link_field()</CODE> calls copy
1807	dynamic-buffer sizes. If the <CODE>O_STATIC</CODE> option is set on one of a
1808	collection of links, buffer resizing will occur only when the field is
1809	edited through that link.
1810	<LI>The call <CODE>field_info()</CODE> will retrieve the original static size of
1811	the field; use <CODE>dynamic_field_info()</CODE> to get the actual dynamic size.
1812	</UL>
1813
1814	<H2><A NAME="fvalidation">Field Validation</A></H2>
1815
1816	By default, a field will accept any data that will fit in its input buffer.
1817	However, it is possible to attach a validation type to a field. If you do
1818	this, any attempt to leave the field while it contains data that doesn't
1819	match the validation type will fail. Some validation types also have a
1820	character-validity check for each time a character is entered in the field. <P>
1821
1822	A field's validation check (if any) is not called when
1823	<CODE>set_field_buffer()</CODE> modifies the input buffer, nor when that buffer
1824	is changed through a linked field. <P>
1825
1826	The <CODE>form</CODE> library provides a rich set of pre-defined validation
1827	types, and gives you the capability to define custom ones of your own. You
1828	can examine and change field validation attributes with the following
1829	functions:
1830
1831	<PRE>
1832	int set_field_type(FIELD field, / field to alter */
1833	FIELDTYPE ftype, / type to associate */
1834	...); /* additional arguments*/
1835
1836	FIELDTYPE field_type(FIELD field); /* field to query */
1837	</PRE>
1838
1839	The validation type of a field is considered an attribute of the field. As
1840	with other field attributes, Also, doing <CODE>set_field_type()</CODE> with a
1841	<CODE>NULL</CODE> field default will change the system default for validation of
1842	newly-created fields. <P>
1843
1844	Here are the pre-defined validation types:
1845
1846	<H3><A NAME="ftype_alpha">TYPE_ALPHA</A></H3>
1847
1848	This field type accepts alphabetic data; no blanks, no digits, no special
1849	characters (this is checked at character-entry time). It is set up with:
1850
1851	<PRE>
1852	int set_field_type(FIELD field, / field to alter */
1853	TYPE_ALPHA, /* type to associate */
1854	int width); /* maximum width of field */
1855	</PRE>
1856
1857	The <CODE>width</CODE> argument sets a minimum width of data. Typically
1858	you'll want to set this to the field width; if it's greater than the
1859	field width, the validation check will always fail. A minimum width
1860	of zero makes field completion optional.
1861
1862	<H3><A NAME="ftype_alnum">TYPE_ALNUM</A></H3>
1863
1864	This field type accepts alphabetic data and digits; no blanks, no special
1865	characters (this is checked at character-entry time). It is set up with:
1866
1867	<PRE>
1868	int set_field_type(FIELD field, / field to alter */
1869	TYPE_ALNUM, /* type to associate */
1870	int width); /* maximum width of field */
1871	</PRE>
1872
1873	The <CODE>width</CODE> argument sets a minimum width of data. As with
1874	TYPE_ALPHA, typically you'll want to set this to the field width; if it's
1875	greater than the field width, the validation check will always fail. A
1876	minimum width of zero makes field completion optional.
1877
1878	<H3><A NAME="ftype_enum">TYPE_ENUM</A></H3>
1879
1880	This type allows you to restrict a field's values to be among a specified
1881	set of string values (for example, the two-letter postal codes for U.S.
1882	states). It is set up with:
1883
1884	<PRE>
1885	int set_field_type(FIELD field, / field to alter */
1886	TYPE_ENUM, /* type to associate */
1887	char *valuelist; / list of possible values */
1888	int checkcase; /* case-sensitive? */
1889	int checkunique); /* must specify uniquely? */
1890	</PRE>
1891
1892	The <CODE>valuelist</CODE> parameter must point at a NULL-terminated list of
1893	valid strings. The <CODE>checkcase</CODE> argument, if true, makes comparison
1894	with the string case-sensitive. <P>
1895
1896	When the user exits a TYPE_ENUM field, the validation procedure tries to
1897	complete the data in the buffer to a valid entry. If a complete choice string
1898	has been entered, it is of course valid. But it is also possible to enter a
1899	prefix of a valid string and have it completed for you. <P>
1900
1901	By default, if you enter such a prefix and it matches more than one value
1902	in the string list, the prefix will be completed to the first matching
1903	value. But the <CODE>checkunique</CODE> argument, if true, requires prefix
1904	matches to be unique in order to be valid. <P>
1905
1906	The <CODE>REQ_NEXT_CHOICE</CODE> and <CODE>REQ_PREV_CHOICE</CODE> input requests
1907	can be particularly useful with these fields.
1908
1909	<H3><A NAME="ftype_integer">TYPE_INTEGER</A></H3>
1910
1911	This field type accepts an integer. It is set up as follows:
1912
1913	<PRE>
1914	int set_field_type(FIELD field, / field to alter */
1915	TYPE_INTEGER, /* type to associate */
1916	int padding, /* # places to zero-pad to */
1917	int vmin, int vmax); /* valid range */
1918	</PRE>
1919
1920	Valid characters consist of an optional leading minus and digits.
1921	The range check is performed on exit. If the range maximum is less
1922	than or equal to the minimum, the range is ignored. <P>
1923
1924	If the value passes its range check, it is padded with as many leading
1925	zero digits as necessary to meet the padding argument. <P>
1926
1927	A <CODE>TYPE_INTEGER</CODE> value buffer can conveniently be interpreted
1928	with the C library function <CODE>atoi(3)</CODE>.
1929
1930	<H3><A NAME="ftype_numeric">TYPE_NUMERIC</A></H3>
1931
1932	This field type accepts a decimal number. It is set up as follows:
1933
1934	<PRE>
1935	int set_field_type(FIELD field, / field to alter */
1936	TYPE_NUMERIC, /* type to associate */
1937	int padding, /* # places of precision */
1938	double vmin, double vmax); /* valid range */
1939	</PRE>
1940
1941	Valid characters consist of an optional leading minus and digits. possibly
1942	including a decimal point. If your system supports locale's, the decimal point
1943	character used must be the one defined by your locale. The range check is
1944	performed on exit. If the range maximum is less than or equal to the minimum,
1945	the range is ignored. <P>
1946
1947	If the value passes its range check, it is padded with as many trailing
1948	zero digits as necessary to meet the padding argument. <P>
1949
1950	A <CODE>TYPE_NUMERIC</CODE> value buffer can conveniently be interpreted
1951	with the C library function <CODE>atof(3)</CODE>.
1952
1953	<H3><A NAME="ftype_regexp">TYPE_REGEXP</A></H3>
1954
1955	This field type accepts data matching a regular expression. It is set up
1956	as follows:
1957
1958	<PRE>
1959	int set_field_type(FIELD field, / field to alter */
1960	TYPE_REGEXP, /* type to associate */
1961	char regexp); / expression to match */
1962	</PRE>
1963
1964	The syntax for regular expressions is that of <CODE>regcomp(3)</CODE>.
1965	The check for regular-expression match is performed on exit.
1966
1967	<H2><A NAME="fbuffer">Direct Field Buffer Manipulation</A></H2>
1968
1969	The chief attribute of a field is its buffer contents. When a form has
1970	been completed, your application usually needs to know the state of each
1971	field buffer. You can find this out with:
1972
1973	<PRE>
1974	char field_buffer(FIELD field, /* field to query */
1975	int bufindex); /* number of buffer to query */
1976	</PRE>
1977
1978	Normally, the state of the zero-numbered buffer for each field is set by
1979	the user's editing actions on that field. It's sometimes useful to be able
1980	to set the value of the zero-numbered (or some other) buffer from your
1981	application:
1982
1983	<PRE>
1984	int set_field_buffer(FIELD field, / field to alter */
1985	int bufindex, /* number of buffer to alter */
1986	char value); / string value to set */
1987	</PRE>
1988
1989	If the field is not large enough and cannot be resized to a sufficiently
1990	large size to contain the specified value, the value will be truncated
1991	to fit. <P>
1992
1993	Calling <CODE>field_buffer()</CODE> with a null field pointer will raise an
1994	error. Calling <CODE>field_buffer()</CODE> on a field not currently selected
1995	for input will return a correct value. Calling <CODE>field_buffer()</CODE> on a
1996	field that is currently selected for input may not necessarily give a
1997	correct field buffer value, because entered data isn't necessarily copied to
1998	buffer zero before the exit validation check.
1999
2000	To guarantee that the returned buffer value reflects on-screen reality,
2001	call <CODE>field_buffer()</CODE> either (1) in the field's exit validation
2002	check routine, (2) from the field's or form's initialization or termination
2003	hooks, or (3) just after a <CODE>REQ_VALIDATION</CODE> request has been processed
2004	by the forms driver.
2005
2006	<H2><A NAME="formattrs">Attributes of Forms</A></H2>
2007
2008	As with field attributes, form attributes inherit a default from a
2009	system default form structure. These defaults can be queried or set by
2010	of these functions using a form-pointer argument of <CODE>NULL</CODE>. <P>
2011
2012	The principal attribute of a form is its field list. You can query
2013	and change this list with:
2014
2015	<PRE>
2016	int set_form_fields(FORM form, / form to alter */
2017	FIELD *fields); / fields to connect */
2018
2019	char form_fields(FORM form); /* fetch fields of form */
2020
2021	int field_count(FORM form); / count connect fields */
2022	</PRE>
2023
2024	The second argument of <CODE>set_form_fields()</CODE> may be a
2025	NULL-terminated field pointer array like the one required by
2026	<CODE>new_form()</CODE>. In that case, the old fields of the form are
2027	disconnected but not freed (and eligible to be connected to other
2028	forms), then the new fields are connected. <P>
2029
2030	It may also be null, in which case the old fields are disconnected
2031	(and not freed) but no new ones are connected. <P>
2032
2033	The <CODE>field_count()</CODE> function simply counts the number of fields
2034	connected to a given from. It returns -1 if the form-pointer argument
2035	is NULL.
2036
2037	<H2><A NAME="fdisplay">Control of Form Display</A></H2>
2038
2039	In the overview section, you saw that to display a form you normally
2040	start by defining its size (and fields), posting it, and refreshing
2041	the screen. There is an hidden step before posting, which is the
2042	association of the form with a frame window (actually, a pair of
2043	windows) within which it will be displayed. By default, the forms
2044	library associates every form with the full-screen window
2045	<CODE>stdscr</CODE>. <P>
2046
2047	By making this step explicit, you can associate a form with a declared
2048	frame window on your screen display. This can be useful if you want to
2049	adapt the form display to different screen sizes, dynamically tile
2050	forms on the screen, or use a form as part of an interface layout
2051	managed by <A HREF="#panels">panels</A>. <P>
2052
2053	The two windows associated with each form have the same functions as
2054	their analogues in the <A HREF="#menu">menu library</A>. Both these
2055	windows are painted when the form is posted and erased when the form
2056	is unposted. <P>
2057
2058	The outer or frame window is not otherwise touched by the form
2059	routines. It exists so the programmer can associate a title, a
2060	border, or perhaps help text with the form and have it properly
2061	refreshed or erased at post/unpost time. The inner window or subwindow
2062	is where the current form page is actually displayed. <P>
2063
2064	In order to declare your own frame window for a form, you'll need to
2065	know the size of the form's bounding rectangle. You can get this
2066	information with:
2067
2068	<PRE>
2069	int scale_form(FORM form, / form to query */
2070	int rows, / form rows */
2071	int cols); / form cols */
2072	</PRE>
2073
2074	The form dimensions are passed back in the locations pointed to by
2075	the arguments. Once you have this information, you can use it to
2076	declare of windows, then use one of these functions:
2077
2078	<PRE>
2079	int set_form_win(FORM form, / form to alter */
2080	WINDOW win); / frame window to connect */
2081
2082	WINDOW form_win(FORM form); /* fetch frame window of form */
2083
2084	int set_form_sub(FORM form, / form to alter */
2085	WINDOW win); / form subwindow to connect */
2086
2087	WINDOW form_sub(FORM form); /* fetch form subwindow of form */
2088	</PRE>
2089
2090	Note that curses operations, including <CODE>refresh()</CODE>, on the form,
2091	should be done on the frame window, not the form subwindow. <P>
2092
2093	It is possible to check from your application whether all of a
2094	scrollable field is actually displayed within the menu subwindow. Use
2095	these functions:
2096
2097	<PRE>
2098	int data_ahead(FORM form); / form to be queried */
2099
2100	int data_behind(FORM form); / form to be queried */
2101	</PRE>
2102
2103	The function <CODE>data_ahead()</CODE> returns TRUE if (a) the current
2104	field is one-line and has undisplayed data off to the right, (b) the current
2105	field is multi-line and there is data off-screen below it. <P>
2106
2107	The function <CODE>data_behind()</CODE> returns TRUE if the first (upper
2108	left hand) character position is off-screen (not being displayed). <P>
2109
2110	Finally, there is a function to restore the form window's cursor to the
2111	value expected by the forms driver:
2112
2113	<PRE>
2114	int pos_form_cursor(FORM ) / form to be queried */
2115	</PRE>
2116
2117	If your application changes the form window cursor, call this function before
2118	handing control back to the forms driver in order to re-synchronize it.
2119
2120	<H2><A NAME="fdriver">Input Processing in the Forms Driver</A></H2>
2121
2122	The function <CODE>form_driver()</CODE> handles virtualized input requests
2123	for form navigation, editing, and validation requests, just as
2124	<CODE>menu_driver</CODE> does for menus (see the section on <A
2125	HREF="#minput">menu input handling</A>).
2126
2127	<PRE>
2128	int form_driver(FORM form, / form to pass input to */
2129	int request); /* form request code */
2130	</PRE>
2131
2132	Your input virtualization function needs to take input and then convert it
2133	to either an alphanumeric character (which is treated as data to be
2134	entered in the currently-selected field), or a forms processing request. <P>
2135
2136	The forms driver provides hooks (through input-validation and
2137	field-termination functions) with which your application code can check
2138	that the input taken by the driver matched what was expected.
2139
2140	<H3><A NAME="fpage">Page Navigation Requests</A></H3>
2141
2142	These requests cause page-level moves through the form,
2143	triggering display of a new form screen.
2144
2145	<DL>
2146	<DT> <CODE>REQ_NEXT_PAGE</CODE>
2147	<DD> Move to the next form page.
2148	<DT> <CODE>REQ_PREV_PAGE</CODE>
2149	<DD> Move to the previous form page.
2150	<DT> <CODE>REQ_FIRST_PAGE</CODE>
2151	<DD> Move to the first form page.
2152	<DT> <CODE>REQ_LAST_PAGE</CODE>
2153	<DD> Move to the last form page.
2154	</DL>
2155
2156	These requests treat the list as cyclic; that is, <CODE>REQ_NEXT_PAGE</CODE>
2157	from the last page goes to the first, and <CODE>REQ_PREV_PAGE</CODE> from
2158	the first page goes to the last.
2159
2160	<H3><A NAME="#ffield">Inter-Field Navigation Requests</A></H3>
2161
2162	These requests handle navigation between fields on the same page.
2163
2164	<DL>
2165	<DT> <CODE>REQ_NEXT_FIELD</CODE>
2166	<DD> Move to next field.
2167	<DT> <CODE>REQ_PREV_FIELD</CODE>
2168	<DD> Move to previous field.
2169	<DT> <CODE>REQ_FIRST_FIELD</CODE>
2170	<DD> Move to the first field.
2171	<DT> <CODE>REQ_LAST_FIELD</CODE>
2172	<DD> Move to the last field.
2173	<DT> <CODE>REQ_SNEXT_FIELD</CODE>
2174	<DD> Move to sorted next field.
2175	<DT> <CODE>REQ_SPREV_FIELD</CODE>
2176	<DD> Move to sorted previous field.
2177	<DT> <CODE>REQ_SFIRST_FIELD</CODE>
2178	<DD> Move to the sorted first field.
2179	<DT> <CODE>REQ_SLAST_FIELD</CODE>
2180	<DD> Move to the sorted last field.
2181	<DT> <CODE>REQ_LEFT_FIELD</CODE>
2182	<DD> Move left to field.
2183	<DT> <CODE>REQ_RIGHT_FIELD</CODE>
2184	<DD> Move right to field.
2185	<DT> <CODE>REQ_UP_FIELD</CODE>
2186	<DD> Move up to field.
2187	<DT> <CODE>REQ_DOWN_FIELD</CODE>
2188	<DD> Move down to field.
2189	</DL>
2190
2191	These requests treat the list of fields on a page as cyclic; that is,
2192	<CODE>REQ_NEXT_FIELD</CODE> from the last field goes to the first, and
2193	<CODE>REQ_PREV_FIELD</CODE> from the first field goes to the last. The
2194	order of the fields for these (and the <CODE>REQ_FIRST_FIELD</CODE> and
2195	<CODE>REQ_LAST_FIELD</CODE> requests) is simply the order of the field
2196	pointers in the form array (as set up by <CODE>new_form()</CODE> or
2197	<CODE>set_form_fields()</CODE> <P>
2198
2199	It is also possible to traverse the fields as if they had been sorted in
2200	screen-position order, so the sequence goes left-to-right and top-to-bottom.
2201	To do this, use the second group of four sorted-movement requests. <P>
2202
2203	Finally, it is possible to move between fields using visual directions up,
2204	down, right, and left. To accomplish this, use the third group of four
2205	requests. Note, however, that the position of a form for purposes of these
2206	requests is its upper-left corner. <P>
2207
2208	For example, suppose you have a multi-line field B, and two
2209	single-line fields A and C on the same line with B, with A to the left
2210	of B and C to the right of B. A <CODE>REQ_MOVE_RIGHT</CODE> from A will
2211	go to B only if A, B, and C <EM>all</EM> share the same first line;
2212	otherwise it will skip over B to C.
2213
2214	<H3><A NAME="#fifield">Intra-Field Navigation Requests</A></H3>
2215
2216	These requests drive movement of the edit cursor within the currently
2217	selected field.
2218
2219	<DL>
2220	<DT> <CODE>REQ_NEXT_CHAR</CODE>
2221	<DD> Move to next character.
2222	<DT> <CODE>REQ_PREV_CHAR</CODE>
2223	<DD> Move to previous character.
2224	<DT> <CODE>REQ_NEXT_LINE</CODE>
2225	<DD> Move to next line.
2226	<DT> <CODE>REQ_PREV_LINE</CODE>
2227	<DD> Move to previous line.
2228	<DT> <CODE>REQ_NEXT_WORD</CODE>
2229	<DD> Move to next word.
2230	<DT> <CODE>REQ_PREV_WORD</CODE>
2231	<DD> Move to previous word.
2232	<DT> <CODE>REQ_BEG_FIELD</CODE>
2233	<DD> Move to beginning of field.
2234	<DT> <CODE>REQ_END_FIELD</CODE>
2235	<DD> Move to end of field.
2236	<DT> <CODE>REQ_BEG_LINE</CODE>
2237	<DD> Move to beginning of line.
2238	<DT> <CODE>REQ_END_LINE</CODE>
2239	<DD> Move to end of line.
2240	<DT> <CODE>REQ_LEFT_CHAR</CODE>
2241	<DD> Move left in field.
2242	<DT> <CODE>REQ_RIGHT_CHAR</CODE>
2243	<DD> Move right in field.
2244	<DT> <CODE>REQ_UP_CHAR</CODE>
2245	<DD> Move up in field.
2246	<DT> <CODE>REQ_DOWN_CHAR</CODE>
2247	<DD> Move down in field.
2248	</DL>
2249
2250	Each <EM>word</EM> is separated from the previous and next characters
2251	by whitespace. The commands to move to beginning and end of line or field
2252	look for the first or last non-pad character in their ranges.
2253
2254	<H3><A NAME="fscroll">Scrolling Requests</A></H3>
2255
2256	Fields that are dynamic and have grown and fields explicitly created
2257	with offscreen rows are scrollable. One-line fields scroll horizontally;
2258	multi-line fields scroll vertically. Most scrolling is triggered by
2259	editing and intra-field movement (the library scrolls the field to keep the
2260	cursor visible). It is possible to explicitly request scrolling with the
2261	following requests:
2262
2263	<DL>
2264	<DT> <CODE>REQ_SCR_FLINE</CODE>
2265	<DD> Scroll vertically forward a line.
2266	<DT> <CODE>REQ_SCR_BLINE</CODE>
2267	<DD> Scroll vertically backward a line.
2268	<DT> <CODE>REQ_SCR_FPAGE</CODE>
2269	<DD> Scroll vertically forward a page.
2270	<DT> <CODE>REQ_SCR_BPAGE</CODE>
2271	<DD> Scroll vertically backward a page.
2272	<DT> <CODE>REQ_SCR_FHPAGE</CODE>
2273	<DD> Scroll vertically forward half a page.
2274	<DT> <CODE>REQ_SCR_BHPAGE</CODE>
2275	<DD> Scroll vertically backward half a page.
2276	<DT> <CODE>REQ_SCR_FCHAR</CODE>
2277	<DD> Scroll horizontally forward a character.
2278	<DT> <CODE>REQ_SCR_BCHAR</CODE>
2279	<DD> Scroll horizontally backward a character.
2280	<DT> <CODE>REQ_SCR_HFLINE</CODE>
2281	<DD> Scroll horizontally one field width forward.
2282	<DT> <CODE>REQ_SCR_HBLINE</CODE>
2283	<DD> Scroll horizontally one field width backward.
2284	<DT> <CODE>REQ_SCR_HFHALF</CODE>
2285	<DD> Scroll horizontally one half field width forward.
2286	<DT> <CODE>REQ_SCR_HBHALF</CODE>
2287	<DD> Scroll horizontally one half field width backward.
2288	</DL>
2289
2290	For scrolling purposes, a <EM>page</EM> of a field is the height
2291	of its visible part.
2292
2293	<H3><A NAME="fedit">Editing Requests</A></H3>
2294
2295	When you pass the forms driver an ASCII character, it is treated as a
2296	request to add the character to the field's data buffer. Whether this
2297	is an insertion or a replacement depends on the field's edit mode
2298	(insertion is the default. <P>
2299
2300	The following requests support editing the field and changing the edit
2301	mode:
2302
2303	<DL>
2304	<DT> <CODE>REQ_INS_MODE</CODE>
2305	<DD> Set insertion mode.
2306	<DT> <CODE>REQ_OVL_MODE</CODE>
2307	<DD> Set overlay mode.
2308	<DT> <CODE>REQ_NEW_LINE</CODE>
2309	<DD> New line request (see below for explanation).
2310	<DT> <CODE>REQ_INS_CHAR</CODE>
2311	<DD> Insert space at character location.
2312	<DT> <CODE>REQ_INS_LINE</CODE>
2313	<DD> Insert blank line at character location.
2314	<DT> <CODE>REQ_DEL_CHAR</CODE>
2315	<DD> Delete character at cursor.
2316	<DT> <CODE>REQ_DEL_PREV</CODE>
2317	<DD> Delete previous word at cursor.
2318	<DT> <CODE>REQ_DEL_LINE</CODE>
2319	<DD> Delete line at cursor.
2320	<DT> <CODE>REQ_DEL_WORD</CODE>
2321	<DD> Delete word at cursor.
2322	<DT> <CODE>REQ_CLR_EOL</CODE>
2323	<DD> Clear to end of line.
2324	<DT> <CODE>REQ_CLR_EOF</CODE>
2325	<DD> Clear to end of field.
2326	<DT> <CODE>REQ_CLEAR_FIELD</CODE>
2327	<DD> Clear entire field.
2328	</DL>
2329
2330	The behavior of the <CODE>REQ_NEW_LINE</CODE> and <CODE>REQ_DEL_PREV</CODE> requests
2331	is complicated and partly controlled by a pair of forms options.
2332	The special cases are triggered when the cursor is at the beginning of
2333	a field, or on the last line of the field. <P>
2334
2335	First, we consider <CODE>REQ_NEW_LINE</CODE>: <P>
2336
2337	The normal behavior of <CODE>REQ_NEW_LINE</CODE> in insert mode is to break the
2338	current line at the position of the edit cursor, inserting the portion of
2339	the current line after the cursor as a new line following the current
2340	and moving the cursor to the beginning of that new line (you may think
2341	of this as inserting a newline in the field buffer). <P>
2342
2343	The normal behavior of <CODE>REQ_NEW_LINE</CODE> in overlay mode is to clear the
2344	current line from the position of the edit cursor to end of line.
2345	The cursor is then moved to the beginning of the next line. <P>
2346
2347	However, <CODE>REQ_NEW_LINE</CODE> at the beginning of a field, or on the
2348	last line of a field, instead does a <CODE>REQ_NEXT_FIELD</CODE>.
2349	<CODE>O_NL_OVERLOAD</CODE> option is off, this special action is
2350	disabled. <P>
2351
2352	Now, let us consider <CODE>REQ_DEL_PREV</CODE>: <P>
2353
2354	The normal behavior of <CODE>REQ_DEL_PREV</CODE> is to delete the previous
2355	character. If insert mode is on, and the cursor is at the start of a
2356	line, and the text on that line will fit on the previous one, it
2357	instead appends the contents of the current line to the previous one
2358	and deletes the current line (you may think of this as deleting a
2359	newline from the field buffer). <P>
2360
2361	However, <CODE>REQ_DEL_PREV</CODE> at the beginning of a field is instead
2362	treated as a <CODE>REQ_PREV_FIELD</CODE>. <P> If the
2363	<CODE>O_BS_OVERLOAD</CODE> option is off, this special action is
2364	disabled and the forms driver just returns <CODE>E_REQUEST_DENIED</CODE>. <P>
2365
2366	See <A HREF="#frmoptions">Form Options</A> for discussion of how to set
2367	and clear the overload options.
2368
2369	<H3><A NAME="forder">Order Requests</A></H3>
2370
2371	If the type of your field is ordered, and has associated functions
2372	for getting the next and previous values of the type from a given value,
2373	there are requests that can fetch that value into the field buffer:
2374
2375	<DL>
2376	<DT> <CODE>REQ_NEXT_CHOICE</CODE>
2377	<DD> Place the successor value of the current value in the buffer.
2378	<DT> <CODE>REQ_PREV_CHOICE</CODE>
2379	<DD> Place the predecessor value of the current value in the buffer.
2380	</DL>
2381
2382	Of the built-in field types, only <CODE>TYPE_ENUM</CODE> has built-in successor
2383	and predecessor functions. When you define a field type of your own
2384	(see <A HREF="#fcustom">Custom Validation Types</A>), you can associate
2385	our own ordering functions.
2386
2387	<H3><A NAME="fappcmds">Application Commands</A></H3>
2388
2389	Form requests are represented as integers above the <CODE>curses</CODE> value
2390	greater than <CODE>KEY_MAX</CODE> and less than or equal to the constant
2391	<CODE>MAX_COMMAND</CODE>. If your input-virtualization routine returns a
2392	value above <CODE>MAX_COMMAND</CODE>, the forms driver will ignore it.
2393
2394	<H2><A NAME="fhooks">Field Change Hooks</A></H2>
2395
2396	It is possible to set function hooks to be executed whenever the
2397	current field or form changes. Here are the functions that support this:
2398
2399	<PRE>
2400	typedef void (HOOK)(); / pointer to function returning void */
2401
2402	int set_form_init(FORM form, / form to alter */
2403	HOOK hook); /* initialization hook */
2404
2405	HOOK form_init(FORM form); / form to query */
2406
2407	int set_form_term(FORM form, / form to alter */
2408	HOOK hook); /* termination hook */
2409
2410	HOOK form_term(FORM form); / form to query */
2411
2412	int set_field_init(FORM form, / form to alter */
2413	HOOK hook); /* initialization hook */
2414
2415	HOOK field_init(FORM form); / form to query */
2416
2417	int set_field_term(FORM form, / form to alter */
2418	HOOK hook); /* termination hook */
2419
2420	HOOK field_term(FORM form); / form to query */
2421	</PRE>
2422
2423	These functions allow you to either set or query four different hooks.
2424	In each of the set functions, the second argument should be the
2425	address of a hook function. These functions differ only in the timing
2426	of the hook call.
2427
2428	<DL>
2429	<DT> form_init
2430	<DD> This hook is called when the form is posted; also, just after
2431	each page change operation.
2432	<DT> field_init
2433	<DD> This hook is called when the form is posted; also, just after
2434	each field change
2435	<DT> field_term
2436	<DD> This hook is called just after field validation; that is, just before
2437	the field is altered. It is also called when the form is unposted.
2438	<DT> form_term
2439	<DD> This hook is called when the form is unposted; also, just before
2440	each page change operation.
2441	</DL>
2442
2443	Calls to these hooks may be triggered
2444	<OL>
2445	<LI>When user editing requests are processed by the forms driver
2446	<LI>When the current page is changed by <CODE>set_current_field()</CODE> call
2447	<LI>When the current field is changed by a <CODE>set_form_page()</CODE> call
2448	</OL>
2449
2450	See <A NAME="ffocus">Field Change Commands</A> for discussion of the latter
2451	two cases. <P>
2452
2453	You can set a default hook for all fields by passing one of the set functions
2454	a NULL first argument. <P>
2455
2456	You can disable any of these hooks by (re)setting them to NULL, the default
2457	value.
2458
2459	<H2><A HREF="#ffocus">Field Change Commands</A></H2>
2460
2461	Normally, navigation through the form will be driven by the user's
2462	input requests. But sometimes it is useful to be able to move the
2463	focus for editing and viewing under control of your application, or
2464	ask which field it currently is in. The following functions help you
2465	accomplish this:
2466
2467	<PRE>
2468	int set_current_field(FORM form, / form to alter */
2469	FIELD field); / field to shift to */
2470
2471	FIELD current_field(FORM form); /* form to query */
2472
2473	int field_index(FORM form, / form to query */
2474	FIELD field); / field to get index of */
2475	</PRE>
2476
2477	The function <CODE>field_index()</CODE> returns the index of the given field
2478	in the given form's field array (the array passed to <CODE>new_form()</CODE> or
2479	<CODE>set_form_fields()</CODE>). <P>
2480
2481	The initial current field of a form is the first active field on the
2482	first page. The function <CODE>set_form_fields()</CODE> resets this.<P>
2483
2484	It is also possible to move around by pages.
2485
2486	<PRE>
2487	int set_form_page(FORM form, / form to alter */
2488	int page); /* page to go to (0-origin) */
2489
2490	int form_page(FORM form); / return form's current page */
2491	</PRE>
2492
2493	The initial page of a newly-created form is 0. The function
2494	<CODE>set_form_fields()</CODE> resets this.
2495
2496	<H2><A NAME="frmoptions">Form Options</A></H2>
2497
2498	Like fields, forms may have control option bits. They can be changed
2499	or queried with these functions:
2500
2501	<PRE>
2502	int set_form_opts(FORM form, / form to alter */
2503	int attr); /* attribute to set */
2504
2505	int form_opts_on(FORM form, / form to alter */
2506	int attr); /* attributes to turn on */
2507
2508	int form_opts_off(FORM form, / form to alter */
2509	int attr); /* attributes to turn off */
2510
2511	int form_opts(FORM form); / form to query */
2512	</PRE>
2513
2514	By default, all options are on. Here are the available option bits:
2515
2516	<DL>
2517	<DT> O_NL_OVERLOAD
2518	<DD> Enable overloading of <CODE>REQ_NEW_LINE</CODE> as described in <A
2519	href="#fedit">Editing Requests</A>. The value of this option is
2520	ignored on dynamic fields that have not reached their size limit;
2521	these have no last line, so the circumstances for triggering a
2522	<CODE>REQ_NEXT_FIELD</CODE> never arise.
2523	<DT> O_BS_OVERLOAD
2524	<DD> Enable overloading of <CODE>REQ_DEL_PREV</CODE> as described in
2525	<A href="#fedit">Editing Requests</A>.
2526	</DL>
2527
2528	The option values are bit-masks and can be composed with logical-or in
2529	the obvious way.
2530
2531	<H2><A NAME="fcustom">Custom Validation Types</A></H2>
2532
2533	The <CODE>form</CODE> library gives you the capability to define custom
2534	validation types of your own. Further, the optional additional arguments
2535	of <CODE>set_field_type</CODE> effectively allow you to parameterize validation
2536	types. Most of the complications in the validation-type interface have to
2537	do with the handling of the additional arguments within custom validation
2538	functions.
2539
2540	<H3><A NAME="flinktypes">Union Types</A></H3>
2541
2542	The simplest way to create a custom data type is to compose it from two
2543	preexisting ones:
2544
2545	<PRE>
2546	FIELD link_fieldtype(FIELDTYPE type1,
2547	FIELDTYPE *type2);
2548	</PRE>
2549
2550	This function creates a field type that will accept any of the values
2551	legal for either of its argument field types (which may be either
2552	predefined or programmer-defined).
2553
2554	If a <CODE>set_field_type()</CODE> call later requires arguments, the new
2555	composite type expects all arguments for the first type, than all arguments
2556	for the second. Order functions (see <A HREF="#forder">Order Requests</A>)
2557	associated with the component types will work on the composite; what it does
2558	is check the validation function for the first type, then for the second, to
2559	figure what type the buffer contents should be treated as.
2560
2561	<H3><A NAME="fnewtypes">New Field Types</A></H3>
2562
2563	To create a field type from scratch, you need to specify one or both of the
2564	following things:
2565
2566	<UL>
2567	<LI>A character-validation function, to check each character as it is entered.
2568	<LI>A field-validation function to be applied on exit from the field.
2569	</UL>
2570
2571	Here's how you do that:
2572	<PRE>
2573	typedef int (HOOK)(); / pointer to function returning int */
2574
2575	FIELDTYPE new_fieldtype(HOOK f_validate, / field validator */
2576	HOOK c_validate) /* character validator */
2577
2578
2579	int free_fieldtype(FIELDTYPE ftype); / type to free */
2580	</PRE>
2581
2582	At least one of the arguments of <CODE>new_fieldtype()</CODE> must be
2583	non-NULL. The forms driver will automatically call the new type's
2584	validation functions at appropriate points in processing a field of
2585	the new type. <P>
2586
2587	The function <CODE>free_fieldtype()</CODE> deallocates the argument
2588	fieldtype, freeing all storage associated with it. <P>
2589
2590	Normally, a field validator is called when the user attempts to
2591	leave the field. Its first argument is a field pointer, from which it
2592	can get to field buffer 0 and test it. If the function returns TRUE,
2593	the operation succeeds; if it returns FALSE, the edit cursor stays in
2594	the field. <P>
2595
2596	A character validator gets the character passed in as a first argument.
2597	It too should return TRUE if the character is valid, FALSE otherwise.
2598
2599	<H3><A NAME="fcheckargs">Validation Function Arguments</A></H3>
2600
2601	Your field- and character- validation functions will be passed a
2602	second argument as well. This second argument is the address of a
2603	structure (which we'll call a <EM>pile</EM>) built from any of the
2604	field-type-specific arguments passed to <CODE>set_field_type()</CODE>. If
2605	no such arguments are defined for the field type, this pile pointer
2606	argument will be NULL. <P>
2607
2608	In order to arrange for such arguments to be passed to your validation
2609	functions, you must associate a small set of storage-management functions
2610	with the type. The forms driver will use these to synthesize a pile
2611	from the trailing arguments of each <CODE>set_field_type()</CODE> argument, and
2612	a pointer to the pile will be passed to the validation functions. <P>
2613
2614	Here is how you make the association:
2615
2616	<PRE>
2617	typedef char (PTRHOOK)(); /* pointer to function returning (char ) /
2618	typedef void (VOIDHOOK)(); / pointer to function returning void */
2619
2620	int set_fieldtype_arg(FIELDTYPE type, / type to alter */
2621	PTRHOOK make_str, /* make structure from args */
2622	PTRHOOK copy_str, /* make copy of structure */
2623	VOIDHOOK free_str); /* free structure storage */
2624	</PRE>
2625
2626	Here is how the storage-management hooks are used:
2627
2628	<DL>
2629	<DT> <CODE>make_str</CODE>
2630	<DD> This function is called by <CODE>set_field_type()</CODE>. It gets one
2631	argument, a <CODE>va_list</CODE> of the type-specific arguments passed to
2632	<CODE>set_field_type()</CODE>. It is expected to return a pile pointer to a data
2633	structure that encapsulates those arguments.
2634	<DT> <CODE>copy_str</CODE>
2635	<DD> This function is called by form library functions that allocate new
2636	field instances. It is expected to take a pile pointer, copy the pile
2637	to allocated storage, and return the address of the pile copy.
2638	<DT> <CODE>free_str</CODE>
2639	<DD> This function is called by field- and type-deallocation routines in the
2640	library. It takes a pile pointer argument, and is expected to free the
2641	storage of that pile.
2642	</DL>
2643
2644	The <CODE>make_str</CODE> and <CODE>copy_str</CODE> functions may return NULL to
2645	signal allocation failure. The library routines will that call them will
2646	return error indication when this happens. Thus, your validation functions
2647	should never see a NULL file pointer and need not check specially for it.
2648
2649	<H3><A NAME="fcustorder">Order Functions For Custom Types</A></H3>
2650
2651	Some custom field types are simply ordered in the same well-defined way
2652	that <CODE>TYPE_ENUM</CODE> is. For such types, it is possible to define
2653	successor and predecessor functions to support the <CODE>REQ_NEXT_CHOICE</CODE>
2654	and <CODE>REQ_PREV_CHOICE</CODE> requests. Here's how:
2655
2656	<PRE>
2657	typedef int (INTHOOK)(); / pointer to function returning int */
2658
2659	int set_fieldtype_arg(FIELDTYPE type, / type to alter */
2660	INTHOOK succ, /* get successor value */
2661	INTHOOK pred); /* get predecessor value */
2662	</PRE>
2663
2664	The successor and predecessor arguments will each be passed two arguments;
2665	a field pointer, and a pile pointer (as for the validation functions). They
2666	are expected to use the function <CODE>field_buffer()</CODE> to read the
2667	current value, and <CODE>set_field_buffer()</CODE> on buffer 0 to set the next
2668	or previous value. Either hook may return TRUE to indicate success (a
2669	legal next or previous value was set) or FALSE to indicate failure.
2670
2671	<H3><A NAME="fcustprobs">Avoiding Problems</A></H3>
2672
2673	The interface for defining custom types is complicated and tricky.
2674	Rather than attempting to create a custom type entirely from scratch,
2675	you should start by studying the library source code for whichever of
2676	the pre-defined types seems to be closest to what you want. <P>
2677
2678	Use that code as a model, and evolve it towards what you really want.
2679	You will avoid many problems and annoyances that way. The code
2680	in the <CODE>ncurses</CODE> library has been specifically exempted from
2681	the package copyright to support this. <P>
2682
2683	If your custom type defines order functions, have do something intuitive
2684	with a blank field. A useful convention is to make the successor of a
2685	blank field the types minimum value, and its predecessor the maximum.
2686	</BODY>
2687	</HTML>

Note: See TracBrowser for help on using the repository browser.

Context Navigation

source: trunk/ncurses/doc/html/ncurses-intro.html@ 2645

Download in other formats: