Meeting minutes
conformance requirements
Vlad: we have some musts that shouldn't be normative should we reword?
Garret: Yes I can do that.
Garret: spec update review, added client/encoder markup to conformance requirements throughout. Also updated algorithm's to use the conformance markup.
test planning
Scott: updates on test generation. Adapted woff2 scripts to generate test fonts.
Scott: updates GSUB table to map p -> PASS, f -> FAIL or the other way around. One or the other will be used depending on the test (whether we expect the client to fail or not)
Scott: then scripts that modify the IFT file to change the format to be invalid.
Scott: just have the one test done so far.
Scott: there's a PR out with these changes.
Garret: will try to review today.
Vlad: do you have a timeline for more test development?
Scott: I'm think maybe about 2 months is my goal.
Garret: can't at the moment, but may have time once I'm done the encoder work to help out with the tests.
IFT encoder updates.
encoder updates.
encoder updates. Progresss on the encoder implementation. Updated to take advantage of codepoitn pair frequencies. Uses a cost function to guide segementation decisions. Cost function is the expected number of bytes transferred. I've been testing against the CJK fonts in the IFT demo and so far have seen 15% reduction in transfer sizes vs
the previous heuristic based approach.
Garret: also implement basic script handling which works for mostly disjoint scripts but does not yet do anything for overlapping scripts.
Skef: an idea for overlapping scripts is to treat them as a pseudo scripts which is configured at a higher level.
Skef: we probably also want processing ahead of segmentation that identifies scripts/languages the font support and only segment based on those.
Garret: good idea, right now the segmenter takes as an input a description of what scripts it should segment across.
Garret: for the frequency data still waiting for approval, hopefully will be ready to release soon.
TPAC meeting.
Vlad: the meeting timing in Japan will be late Sunday which is inconvenient for remote participants.
Vlad: planned to start 9am - 5pm Japan time.
Vlad: which is roughly 8pm eastern
Vlad: 5pm pacific.
Vlad: we can move the start time earlier.
Vlad: and plan to wrap up in ~4 hours.
Skef: since this is the first day of the conference may be problematic to move earlier.
Vlad: will check with conference organizers about moving time and/or day.
Skef: we should try and get some time with CSS folks to discuss javascript API for IFT>
Vlad: would be best to get an official block booked for that, will add that to my request.
next call schedule.
Vlad: consider 4th of November to have one last meeting prior to TPAC.
Vlad: for TPAC propose that main discussion topics will be around development/design of the testing plan.
Garret: sounds good.
Vlad: would be good to have a video demo like Chris mentioned.
Garret: I can get that figured out.