Meeting minutes
handling CFF and CFF2 (w3c/IFT#260 )
Due to higher complexity of CFF/CFF2 tables we decided we need to introduce some CFF specific requiremetns into the spec:
- IFT mapping table stores an offset to the start of charstrings so client can locate it without parsing CFF.
- Charstrings table is the last piece of data in CFF so it can be extended without modifying any of the rest of the table.
Implementation updates:
- client support for CFF and CFF2 including changing offset sizes.
- encoder support for CFF, CFF2 coming soon.
Garret: shouldn't be any additional requirements to add to the spec around CFF/CFF2 handling.
Skef: I agree.
URI templates (w3c/IFT#259 )
Garret: summary of this, in the spec replaced specification for URLs with whatwg URL which is the more modern one. Limited template syntax to level 1 and disallowed undefined variables. This significantly simplifies implementation of a template expander. I also implemented a proof of concept one in the client to see how it would look.
Garret: now waiting to here from annevk on the github issue if further concerns remain.
review of remaining open issues.
Garret: 210 planning to get back to this soon.
Skef: 201 serves as a remainder to keep this in mind. Probably want to keep it open until we get to the next stage.
Vlad: next stage spec will be more frozen so seems like a good point.
Vlad: 194, 193 tied to wide review will be closed once that finishes.
Skef: 192 can likely be closed.
Garret: agreed, closed it.
Garret: 125, client conformance tests.
Vlad: once spec is complete won't be able to progress further until we have conformance tests around the "must"s in the spec.
Vlad: we need ideas and implementations for each one.
Vlad: to give something to client to have it pass or fail.
Vlad: and record the test result.
Vlad: woff2 tests could be a good reference
Vlad: wonder if Scott would be able to take a look at this?
Scott: In about two weeks I should have more time to contribute.
Vlad: for now could just be prep work to help the development of the tests later.
Skef: for conformance tests we may also need to support non web browser clients since thats a goal of the spec.
Garret: agreed, may want to have tests output both as web pages but also in another form (eg. set of font files and expected outcomes) that non browser clients could use to test their implementations.
Garret: #109 recommend we close for now, and ask Yoav to reopen with more details if their are still concerns.
Garret: 101, still seeing steady progress on the shared brotli spec. Last change was in feb, will continue to trakc.k