Define what block percentages resolve against.

[bfgeek]

https://www.software.hixie.ch/utilities/js/live-dom-viewer/?saved=6703

Safari appears to follow the quirks mode behaviour of walking up ancestor chain.
Firefox appears to resolve against zero.

This needs to be defined, with tests, and various writing modes etc.

[bfgeek]

My preferred mode would be to follow the new grid/flex/layout-api behaviour of not passing parents percent resolution size down to children.

However this also depends on what is decided for the width/height issue.

[fred-wang]

I'll put this on the agenda of the next MathML CG meeting.
See https://github.com/mathml-refresh/mathml/issues/45#issuecomment-508735440

I think it does not make much sense for math layout to resolve percent size for children as relative to the parent size (unless maybe int table cells?). That complicates things and is likely to be even worse with stretchy operators.

I see two interpretations possible:
(1) Like mspace@width and mspace@height in MathML 3: Relative to the default attribute value, which is 0.
(2) Like mpadded@width and mpadded@height in MathML 3: Relative to the content's size.

Since mspace content is empty, (2) would also work for mspace.

[fred-wang]

Consensus during mathml core meeting: Ignore width/height on all new layout as they are no clear use case and that simplifies implementation. It still makes sense for mtable or math elements.

[bfgeek]

You still need to define this for nested elements which do abide by width/height. (See test-case).

[fred-wang]

You still need to define this for nested elements which do abide by width/height. (See test-case).

Proposal: resolve percent against 0.

[NSoiffer]

This was discussed (argued :-) in the 9/9/19 MathML core meeting. In that meeting, @fred-wang argued for '0' and @NSoiffer argued for the containing block (which could be mtable or math)

The argument for '0' is simplicity of implementation. The argument for the containing block is consistency with HTML. @rwlbuis was going to look into how hard it really would be to look up and use the containing block's render width/height.

[fred-wang]

Consensus from 2019/09/12 is essentially: use the size provided by a block container without display: math

[davidcarlisle]

Still needs spec change (spec still has issue link note)