[MPEG-OTSPEC] VARC issues with the published OFF 5th ed

[Behdad Esfahbod]

Hello everyone,

Thank you, Vlad, and everyone else who worked in the past four years to get
the boring-expansion proposals into the ISO OFF 5th ed.

Since the PDF became freely available on Monday, I've been busy
cross-checking my implementations against the spec. Generally things are
holding up well, but unfortunately I discovered three issues in the VARC
table that I recommend fixing in the spec. Details and recommendation in
the github issues:


*1. VARC ConditionList: restore omitted uint32 conditionCount*
This one is was lost in translation from my working design to the ISO
proposal. It's currently impossible to implement the ISO spec, because it
defines an array without defining its length anywhere:

  https://github.com/harfbuzz/boring-expansion-spec/issues/176

*2. VARC: make SparseVariationRegion axis offsets table-relative and
correct its designation*

This was was a mistake from the review modifications of the table format to
promote a struct to a table, but this was not done properly. Offsets should
be relative to the beginning of their table, not the grandparent table as
currently speced.

  https://github.com/harfbuzz/boring-expansion-spec/issues/177

*3. VARC: Remove the exception to ordinary LSB/origin alignment for
components*

This is a design leftover from when I had encoded the variable-composites
in the `glyf` table. In that design, the recommendation made sense. In a
separate `VARC` table, it doesn't, and hinders implementation. This change
wouldn't affect properly-made fonts (lsb=xMin).

  https://github.com/harfbuzz/boring-expansion-spec/issues/173


That's it for now. I will be spending a lot more time with the spec, but I
don't expect other major issues to arise.

Thanks,
behdad
http://behdad.org/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.aau.at/pipermail/mpeg-otspec/attachments/20261002/eacefd7b/attachment.htm>