Backport to 5.0.0

[LaurenzV]

Status	Commit message	HB
🟢	[tt-font] Apply VVAR.vOrg variation to VORG origin	Link	Implemented in ttf-parser. I manually verified that https://github.com/harfbuzz/harfbuzz/issues/1694 is now resolved for us as well.
🟡	[glyf] Change side-bearing rounding	Link	not relevant I think? since we don't support phantom points yet
🟡	Fix LookupFlag negation	Link	I believe this shoulddnt affect us?
🟡	[use-table] Port to using packtab	Link	Unfortunately packtab seems to generate c++ code (and it's also been updated in the meanwhile)... so for now I've decided to switch to a manual implementation, hopegully when the port is done we can port the packTab code to work with rust. But see the discussion below.
🟢	[GPOS] Adjust mark attachment on multiple substitution some more	Link
🟡	[reorg] Move Coverage	Link	ttf-parser stuff I guess?
🟢	[arabic] Pause after calt only if no rclt	Link
🟡	[aat/morx] Implement language-specific forms	Link	Where exactly can I find this ltag in our port?
🟡	[aat/morx] Add test for previous commit	Link
🟡	Add hb_language_matches()	Link
🟢	[arabic] Make more features F_MANUAL_ZWJ	Link
🟢	[GPOS/Cursive] Fix breaking of parent-child attachment	Link
🟢	[hebrew] Implement Jerusalem mark reordering	Link
🟢	[hebrew] Comment	Link
🟢	Add tests for multiple cursive positioning	Link
🟢	[hebrew] Break out of reordering loop when pattern found	Link
🟢	[GPOS/CursivePos] Fix unsafe-to-break marking	Link
🟡	[layout] Reduce number of closure rounds	Link	we dont have this right?

[RazrFalcon]

How the new USE table works? Have you simply ported C code to Rust? Because the new Python script produces C, right?

[LaurenzV]

Yeah exactly, they have a python package that somehow encodes the data more compactly, and it generates C code. I think we can rewrite it to produce Rust, but I noticed that they change the script again in the future so for now I think it's easier to do it by hand, it only changes a couple of times anyway.

[behdad]

Adding Rust output to packtab sounds like it should take 20min. I'd be happy to take that.

https://github.com/harfbuzz/packtab/blob/3eb74c28bc92379156156234fa7d11d63e27cd75/Lib/packTab/__init__.py#L194

[LaurenzV]

That would be great if possible!

[LaurenzV]

@RazrFalcon Alright, so the main thing missing now is avar2. Would you prefer me to try implement it in ttf-parser as suggested by Behdad, or should I rather focus on backporting more commits for now?

[LaurenzV]

And I've added a note about the ones I was uncertain about again at the top.

[RazrFalcon]

I think we can skip avar2 for now. We could mention it in the README, that this harfbuzz feature is not supported.

To my understanding avar2 is not part of the TrueType spec. It might become one, but not for now. So we can ignore it as an extreme edge case. We have other things to do in the mean time...

Of course if you're interested in implementing avar2 - go ahead.

[LaurenzV]

I'll add it in the README.

[LaurenzV]

Okay, so apart from the orange points mentioned above that I was unsure about, the rest should be done.

[LaurenzV]

Closing in favor of #109, which contains all commits here + until 5.3.1.

[RazrFalcon]

Why not merge it as 5.0.0?

[LaurenzV]

I mean... we can if you want, but since I already ported until 5.3.1 might as well merge that, no? The commits from 5.0.0 are in there as well, but if you prefer it that way I will reopen one just for 5.0.0.

[RazrFalcon]

I guess I would prefer smaller patches in the future. Easier to review. But yeah, I was just confused why have you merged two PRs into one.

[LaurenzV]

Okok, I'll keep that in mind for the future.