Meeting minutes
dropping patch map format 1
Garret: format 1 was added at the start as a more compact but less functional way to encode mappings. After development of the segmenter we've found that its practically unused. So to save spec, and implementation complexity I'm suggesting that we drop it.
Skef: is format 1 less complex for implementations?
Garret: In my opinion is more complex due to the codepoint to glyph indirection and the feature mapping mechanism.
Garret: the format 2 byte was left as an extensibility point and to avoid breaking existing fonts.
Chris: we should add a note about why that is.
Garret: I'll update the PR.
Suzuki-san: not using format 1, so no objection
<ChrisL> +1 to removing patch format 1
RESOLUTION: remove format 1.
mime type discussions.
Garret: I've tested changing both the flavour and signature and found that changing either breaks font use in browsers.
Skef: remaining consideration is that we allow bare fonts to be transferred without woff.
Skef: fonts unecoded on local machine we should think about how we want that to look.
Skef: decided on possibility of font ift for that format.
Skef: different file extension too.
.ift or .itf
Skef: suggested way of storing unencoded file.
Garret: encoder doesn't output raw fonts at the moment.
Chris: what you want is a simple mapping from file name to mime type.
Chris: so in favor of font/ift.
Skef: normal situation is encoder makes woff2 with different flavour
Skef: 2 considerations not downloading what you don't need, and not trying to use something that you can't use.
Chris: third most content creators have no control over how servers are set up.
Skef: so is it a problem that we are introducing inner type in woff2 that things can't use?
Chris: I htink that's fine, its already the case today eg. COLRv1.
Skef: do we need to worry about unecoded fonts?
Garret: I don't think so, encoder won't be producing these.
Skef: if there are legacy impl that decide to woff2 unecode, there's no specialfile extension, and there's a special flavour.
Garret: summary of plan - don't introduce new mime type, don't specify special extension, specify a new flavor value.
Garret: I can work on the spec change.
Skef: why this is important, we're making this new thing. There's lots of existing font licenses. Are the licenses compatible? same question with woff2 because it was not completely lossless. People thought woff2 was compatible because the behaviour doesn't change. Because it's just a wrapper.
Skef: so we started creating these fonts that look like regular fonts with minimal data, then they don't behave like the original fonts.
Skef: concern that people would think it's not a simple container, then worried about licensing.
Skef: trying to build on precedent that woff2 set.
Vlad: subsetting is something that's been around for awhile.
Skef: subsets may not have been testing legally yet.
Chris: early on you had fonts split into separate files.
Skef: those were made by the foundry.
Vlad: people can understand optimization to fonts.
Garret: encoder has been design to maintain all original font behaviour.
Garret: we will still need to register the font/ift-patch
Chris: I'll need to review the font/ mimetype doc, let's come back to this next week.
CTS development.
Garret: good progress is continuing to be made, no issues to discuss on my end.
woff turned sixteen!
Vlad: Chris published blog post about it.
<ChrisL> https://
<ChrisL> also in Chinese https://
<ChrisL> (It would be good to have this in Japanese too)
Garret: we will take a break from calls in august due to travel.