[MPEG-OTSPEC] AHG Zoom meeting summary

Thu Jul 6 19:40:52 CEST 2023

Dear AHG members,

Thank you very much for your time commitment and active participation in
today's meeting. Our discussions covered many topics, including both
technical proposals and procedural issues.

To answer many questions about the process:
Working Draft OFF (5th edition) can be updated based on the decisions made
during WG meetings. These updates can only be based on content submitted as
input contributions, after their review and acceptance that happens as part
of the WG meeting discussions. For as long as the document remains a
Working Draft, additional changes can be easily proposed by new input
contributions for future meetings, and these changes can modify any part of
the spec, regardless of whether it is something that was recently added or
have been part of the previous spec editions.
It has been my intention to make sure that each newly updated version of
the Working Draft should always be made available for public review, and
that the AHG should be able to review and comment on the proposed changes
(preceding the follow up WG discussions), before the changes are made.

A brief summary of [lengthy] technical discussions that were centered on
the content of the "Technologies under consideration" document (
https://www.mpeg.org/wp-content/uploads/mpeg_meetings/142_Antalya/w22631.zip),
the AHG has reached a consensus to:
- Recommend the submission of the proposed changes to 'avar' (version 2) as
a separate input contribution, and recommend adoption of these changes as
part of the next revision of the Working Draft OFF.
-Prioritize further discussion and spec development work related to
overcoming current 16-bit glyph ID limitation and extending the format
capabilities to support large fonts that would include more than 64K
glyphs. This work should be primarily focused on developing a mechanism
that would be backward compatible, would allow new larger fonts be
compatible with older implementations (and essentially behave like
currently supported fonts). Our efforts for the next 6-9 months should be
focused on this approach, until such time if and when the determination is
made on whether achieving this goal is feasible and practical.
- Postpone the work on the remaining parts described by the TuC document,
namely 'glyf' table support for cubic Bezier curves and variable composites
until the work on extending the 64K glyph limit is completed.

Our future planned activities will utilize multiple modes of communication,
including this email list, GitHub discussions (
https://github.com/harfbuzz/boring-expansion-spec/issues), and Zoom &
face-to-face meetings. Our first opportunity to organize the F2F meeting
would be to co-locate the venue of the meeting with the annual TypeCon2023
conference (happening on August 17-19 in Portland, Oregon. Please share
with this AHG your availability and current attendance plans, and please be
aware that the early bird registration rate is still available for the next
2 days: https://www.typecon.com/typecon-registration

Looking forward to hearing from you re: your availability to attend the F2F
AHG meeting co-located with TypeCon2023.

Thank you once again for your active participation,
Vladimir

P.S. The meeting chat content is attached FYI.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.aau.at/mailman/private/mpeg-otspec/attachments/20230706/d02421aa/attachment-0001.html>
-------------- next part --------------
10:02:45 From Vladimir Levantovsky to Everyone:
	ISO Code of Conduct: https://www.iso.org/files/live/sites/isoorg/files/store/en/PUB100011.pdf
10:03:18 From Vladimir Levantovsky to Everyone:
	Declaration for participants: https://www.iso.org/declaration-for-participants-in-iso-activities.html
10:14:09 From Zachary Scheuren to Vladimir Levantovsky(Direct Message):
	It’s late in Japan and family is sleeping nearby so I’m keeping quiet.
10:14:30 From Kalapi Gajjar-Bordawekar to Everyone:
	Hello all! I can’t unmute at this time so I’ll introduce myself here:

	I’m Kalapi. I’m a typeface designer and font engineer from India. I run a small independent type foundry.

	Thanks for having me.
10:16:08 From Zachary Scheuren to Everyone:
	Same as Kalapi. Middle of the night and baby is sleeping nearby so I’m muting. 

	I’m a font developer at Adobe working on CJK and more.
10:16:34 From suzuki toshiya (Hiroshima Univ) to Everyone:
	Reacted to "Hello all! I can’t u..." with 👍
10:16:39 From suzuki toshiya (Hiroshima Univ) to Everyone:
	Reacted to "Same as Kalapi. Midd..." with 👍
10:17:07 From Frank Grießhammer to Everyone:
	Reacted to "Hello all! I can’t u..." with 👍
10:17:08 From Frank Grießhammer to Everyone:
	Reacted to "Same as Kalapi. Midd..." with 👍
10:20:46 From Bianca Berning to Everyone:
	Reacted to "Same as Kalapi. Midd..." with 👍
10:20:46 From Bianca Berning to Everyone:
	Reacted to "Hello all! I can’t u..." with 👍
10:22:12 From Eiso Chan to Everyone:
	Reacted to "Same as Kalapi. Midd…" with 👍
10:22:14 From Eiso Chan to Everyone:
	Reacted to "Hello all! I can’t u…" with 👍
10:22:58 From John Hudson to Everyone:
	Can someone post a link to the document? I don’t have it handy on my new computer.
10:23:43 From Ned Holbrook to Everyone:
	https://www.mpeg.org/wp-content/uploads/mpeg_meetings/142_Antalya/w22631.zip
10:23:51 From John Hudson to Everyone:
	Thanks!
10:23:51 From Frank Grießhammer to Everyone:
	Reacted to "https://www.mpeg.org..." with 👍
10:23:55 From suzuki toshiya (Hiroshima Univ) to Everyone:
	Reacted to "https://www.mpeg.org..." with 👍
10:24:57 From Vladimir Levantovsky to Zachary Scheuren(Direct Message):
	Just to note - the link to the document was also included as part of the meeting invite.
10:30:19 From Kalapi Gajjar-Bordawekar to Everyone:
	Reacted to "Same as Kalapi. Midd…" with 👋
10:31:05 From Ken Lunde to Everyone:
	A full-blown Pan-CJK font implementation that supports all CJK Unified Ideographs blocks in their entirety would completely break what is being discussed due to the necessary complexity of the 'locl' GSUB feature.
10:32:01 From Hin-tak Leung to Everyone:
	Reacted to "A full-blown Pan-CJK..." with 👍
10:32:01 From Rod S to Everyone:
	Ned's hand is up
10:32:09 From Eiso Chan to Everyone:
	Reacted to "A full-blown Pan-CJK…" with 👍
10:34:02 From suzuki toshiya (Hiroshima Univ) to Everyone:
	Reacted to "A full-blown Pan-CJK..." with 👍
10:55:43 From Dave Crossland phone to Everyone:
	Why can't we get that focused engagement again?
10:56:26 From Dave Crossland phone to Everyone:
	Avar3
10:56:38 From Dave Crossland phone to Everyone:
	Where was the money coming from for vf in 2016?
10:59:43 From Cosimo Lupo to Everyone:
	which existing fonts?
10:59:45 From Dave Crossland phone to Everyone:
	The spec should document the implementations in the wild and if Ms has suggestions for avar3 that's fantastic
10:59:56 From Adam Twardoch to Everyone:
	People have implemented COLRv0 (MS), OTVar (Apple & FT), sbix (Apple), OS2v5 (MS), MERG (MS) before even vendor specs were published, let alone submitting to ISO :)
11:00:15 From Dave Crossland phone to Everyone:
	Reacted to "People have implemen..." with 👏
11:00:44 From Adam Twardoch to Everyone:
	Or they've implemented "predecessors" to the standardized stuff
11:01:44 From Dave Crossland phone to Everyone:
	I fundamentally dont see the difference between now and 2016, why can't we recreate the successful environment of 2016?
11:02:00 From Hin-tak Leung to Everyone:
	Reacted to "People have implemen..." with 👍
11:02:10 From Adam Twardoch to Everyone:
	sbix, gvar, fvar implementations were all "post-fixed". Silent implementations exist, obviously, because hands-on implementations unveil problems with "paper specs".
11:02:33 From suzuki toshiya to Everyone:
	Reacted to "sbix, gvar, fvar imp..." with 👍
11:02:41 From Hin-tak Leung to Everyone:
	Reacted to "sbix, gvar, fvar imp..." with 👍
11:05:14 From Adam Twardoch to Everyone:
	These days often the silent implementations end up in opensource software so they are visible (rather than truly silent), which ofc makes some people starting doing prod work early. That's the "beauty" of "continuous" rather than strictly "major-version-discrete" software deployment.
11:06:26 From suzuki toshiya to Everyone:
	Reacted to "These days often the..." with 👍
11:07:48 From Dave Crossland phone to Everyone:
	Reacted to "These days often the..." with 👍
11:20:29 From John Hudson to Everyone:
	Q. Can the group operate in a way that compensates for resource constraints in specific quarters?
11:21:11 From Niklas Borson to Everyone:
	New fonts at least need to print.
11:21:27 From Laurence Penney to Everyone:
	We can in principle add per-glyph avar2 without breaking existing fonts, in the same way the OS/2 is extensible without breaking older readers. For example, we can simply add more offset fields in the avar2 format (with or without a minor version bump), pointing to subtables that have yet to be defined.
11:21:43 From Dave Crossland phone to Everyone:
	Im not sure John, it's Microsoft's goal to keep oT in sync with off and if they aren't resourced to do that themselves, and they won't open their document source repo for others to contribute, I dont see how we can help them
11:22:23 From Rod S to Everyone:
	I agree re printing, we should update the proposal to more directly address
11:26:13 From Laurence Penney to Everyone:
	Reacted to "sbix, gvar, fvar imp..." with 👍
11:27:14 From Adam Twardoch to Everyone:
	"Print" is a specific case of "downgrade". 

	Contour instantiation is one example of downgrade from variable contours to static contours (I myself don't exactly see how avar2 adds more trouble to it over OTVar per se).  

	For >64k, a downgrade strategy is splitting/subsetting. 

	For various color flavors, it's not clear. Vendors downgrade differently: they may rasterize into bitmaps or they may use monochrome contour glyphs. And if SVG+sbix+(glyf+COLR) co-exist in one font, we don't really know what will happen
11:27:43 From Cosimo Lupo to Everyone:
	Reacted to ""Print" is a speci..." with 👍
11:28:50 From Adam Twardoch to Everyone:
	I think a spec (official or not) should give guidelines on how to "downgrade" from various new things in OT — not just the things we intro now but also for things added in the last 10 years :)
11:30:30 From Bianca Berning to Everyone:
	Sorry folks, I have to sign off. Was good to hear where everyone’s at. Thanks for having me.
11:30:46 From suzuki toshiya to Everyone:
	Reacted to "Sorry folks, I have ..." with ☺️
11:30:49 From Cosimo Lupo to Everyone:
	Reacted to "Sorry folks, I hav..." with 👋
11:31:08 From Adam Twardoch to Everyone:
	I'd also like to add that in 2016 we introduced the first truly breaking change since the inception of OT: CFF2 = fonts that fully "fail" in older systems.
11:34:11 From John Hudson to Everyone:
	Further to what Dave is saying, there are variable fonts I have not made because I have lacked the axis model that would make them feasible.
11:34:36 From Adam Twardoch to Everyone:
	Reacted to "Further to what Dave…" with 👍
11:35:01 From Laurence Penney to Everyone:
	Few tests have been done but I would expect avar2 to help improve/shrink CFF2 fonts just as much as it does glyf+gvar fonts.
11:35:17 From Adam Twardoch to Everyone:
	Reacted to "Few tests have been …" with 👍
11:35:38 From Dave Crossland phone to Everyone:
	Reacted to "Further to what Dave..." with ❤️
11:36:26 From John Hudson to Everyone:
	I second Skef’s suggested approach.
11:36:53 From Dave Crossland phone to Everyone:
	Seems slow, vs to call avar2 and move on?
11:37:27 From John Hudson to Everyone:
	Was thinking in terms of the working draft schedule suggested by Vlad.
11:38:03 From John Hudson to Everyone:
	Four colour formats anyone? 🙂
11:38:13 From Niklas Borson to Everyone:
	If we're going to prioritize, it seems like AVAR.Next and 24-bit glyph IDs are both important (to different audiences). Extending the glyf table to support composites and cubic Beziers seem lower priority to me.
11:38:20 From Laurence Penney to Everyone:
	We can add avar offset fields if the Apple & Harfbuzz implementations don’t complain about holes in the table as they understand it, and still call it avar2.
11:38:34 From Dave Crossland phone to Everyone:
	VarCo is essential to our CJK fonts
11:38:45 From suzuki toshiya to Everyone:
	Reacted to "If we're going to pr..." with 👍
11:38:46 From Hin-tak Leung to Everyone:
	Slight side-tracting comment: I think the size-saving aspect of COLRv1 over OT-SVG is somewhat overstated. It is true that COLRv1 fonts are smaller than equivalent"poorly-made" OT-SVG fonts, that's all. ;-p
11:38:56 From Dave Crossland phone to Everyone:
	1. Avar2 2. VarCo 3. B65k 4. Cubic in glyf
11:39:03 From Dave Crossland phone to Everyone:
	Is my priority order
11:40:18 From Niklas Borson to Everyone:
	@Dave Crossland phone, interesting that VarCo is above >64K gid for you. I'd also expect it to be lower cost.
11:40:55 From Zachary Scheuren to Everyone:
	I agree that VarCo is a big deal. That will shrink our CJK fonts by an insane amount.
11:42:41 From Dave Crossland phone to Everyone:
	65k+ gids is important long term (mj+) but font fallback systems work ok
11:42:55 From Niklas Borson to Everyone:
	Reacted to "65k+ gids is importa..." with 👍
11:43:08 From Eiso Chan to Everyone:
	Reacted to "65k+ gids is importa…" with 👍
11:43:22 From suzuki toshiya to Everyone:
	Reacted to "65k+ gids is importa..." with 😢
11:43:24 From Adam Twardoch to Everyone:
	There's also COLR1 right? I mean that's not fully in OFF? Or we're "done" with that? :)
11:43:44 From Hin-tak Leung to Everyone:
	Reacted to "There's also COLR1 r..." with 👍
11:43:53 From Dave Crossland phone to Everyone:
	Typecon!
11:44:55 From Rod S. to Everyone:
	Tentatively yes
11:45:16 From GregH to Everyone:
	GregH: Tentatively yes
11:45:55 From Dave Crossland phone to Everyone:
	Would be great for someone from Adobe to join, with the above then I'll start to organize something
11:47:55 From John Hudson to Everyone:
	Typecon conference dates are 17–19 August. Possibly meet on 16th? Or two days?
11:53:27 From MURATA to Everyone:
	When do we close this meeting?
11:55:47 From Frank Grießhammer to Everyone:
	Reacted to "When do we close thi..." with 👆
11:55:50 From Hin-tak Leung to Everyone:
	Argh, it is going pass midnight by quite a bit in Japan now...
11:56:04 From Cosimo Lupo to Everyone:
	Reacted to "When do we close t..." with 👆
11:56:15 From suzuki toshiya to Everyone:
	Reacted to "Argh, it is going pa..." with 😅
11:57:06 From Zachary Scheuren to Everyone:
	Almost 1am here in Japan and I’ve been up since 5am. Was expecting this to be done an hour ago.
11:58:23 From Frank Grießhammer to Everyone:
	We will bring up Adobe participation at the TypeCon session next week, when the US company holiday is over.
11:59:13 From Laurence Penney to Everyone:
	I wonder if Peter could make an informal draft of Microsoft’s thoughts on per-glyph avar2, so that others might help to develop the idea.
11:59:24 From Ken Lunde to Everyone:
	I am getting kicked out of the room I have been using, so I am signing off.
11:59:43 From Peter Constable to Everyone:
	For my part, I can’t commit to attending a meeting during TypeCon as I’m in the process of moving. Going forward, I would have a strong preference to having face-to-face meetings only in Washington, California or Arizona.
12:01:22 From Dave Crossland phone to Everyone:
	I believe MS has had AMPLE opportunity. It's sad that was not possible to execute on, but I'm not seeing the tragedy of avar3
12:01:49 From Dave Crossland phone to Everyone:
	I think we will need a f2f in Japan
12:01:54 From Dave Crossland phone to Everyone:
	Next year
12:02:03 From Dave Crossland phone to Everyone:
	Reacted to "We will bring up Ado..." with 👍
12:02:50 From suzuki toshiya to Everyone:
	Reacted to "I am getting kicked ..." with 👋
12:04:29 From MURATA to Everyone:
	Reacted to "I think we will need..." with 👍
12:06:48 From Dave Crossland phone to Everyone:
	I have to go, I'll work with Vlad to organize something at typecon
12:07:02 From suzuki toshiya to Everyone:
	Reacted to "I have to go, I'll w..." with 👍
12:07:17 From Laurence Penney to Everyone:
	I also see nothing wrong with incorporating avar as it is now, but further enhancements in a year or in 10 years.
12:08:23 From Rod S. to Everyone:
	Reacted to "I also see nothing..." with 👍
12:10:24 From skef to Everyone:
	I'll check with my management chain about my going to TypeCon
12:11:07 From Rod S. to Everyone:
	Reacted to "I think we will ne..." with 👍