Bug 220536 – Multi-codepoint CJK grapheme clusters are not oriented correctly in vertical writing mode

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Myles C. Maxfield]

Multi-codepoint CJK grapheme clusters are not oriented correctly in vertical writing mode

[Myles C. Maxfield]

Created attachment 417438 [details]
Patch

[Myles C. Maxfield]

<rdar://problem/70633887>

[Myles C. Maxfield]

Investigating the Windows failure now.

[Myles C. Maxfield]

Committed r271425: <https://trac.webkit.org/changeset/271425>

[Lauro Moura]

GTK/WPE renders the glyph correctly but with a slightly different positioning, ditto for Firefox 85 and Chromium 87. The same occurs for single codepoints.

Does the spec require them to to be layout exactly the same (i.e. single grapheme in vertical and horizontal modes) or it's something open and the test could be tweaked to work in GTK/WPE?

[Myles C. Maxfield]

(In reply to Lauro Moura from comment #5)
> GTK/WPE renders the glyph correctly but with a slightly different
> positioning, ditto for Firefox 85 and Chromium 87. The same occurs for
> single codepoints.
> 
> Does the spec require them to to be layout exactly the same (i.e. single
> grapheme in vertical and horizontal modes) or it's something open and the
> test could be tweaked to work in GTK/WPE?

They don’t *have* to be exactly the same. However, the em box is usually a perfect square, so it would be good to at least determine where the offset comes from in GTK/WPE. One would expect that rotating a perfect square by 90deg would lead to an identical rendering.

[Lauro Moura]

(In reply to Myles C. Maxfield from comment #6)
> (In reply to Lauro Moura from comment #5)
> > GTK/WPE renders the glyph correctly but with a slightly different
> > positioning, ditto for Firefox 85 and Chromium 87. The same occurs for
> > single codepoints.
> > 
> > Does the spec require them to to be layout exactly the same (i.e. single
> > grapheme in vertical and horizontal modes) or it's something open and the
> > test could be tweaked to work in GTK/WPE?
> 
> They don’t *have* to be exactly the same. However, the em box is usually a
> perfect square, so it would be good to at least determine where the offset
> comes from in GTK/WPE. One would expect that rotating a perfect square by
> 90deg would lead to an identical rendering.

For GTK/WPE in this test the box seems to be slightly rectangular (111x100) and slightly offset (-6,0 and 0,-6, depending on the orientation). I've opened bug220680 to track it.