Meeting minutes
present!
First discussion: Integrating client state into the client table in the font
There is a PR with the changes
We now have a format for the data
Request and response descriptions are modified accordingly
Response is much simpler
Overall Garret considers the changes positive relative to spec complexity
Possible further changes:
Considering having the client state object attached to the request rather than unpacking and mixing the fields as they are now
Skef agrees that would be a good change
(Might need special handling for the codepoint_ordering field.)
Second topic: Future of VCDiff in spec as a required implementation
Skef worried about the complexity of Brotli, and also the modularity of the implementations
Vlad worried about optional spec parts, favors making Brotli mandatory
Brotli font-aware patch compression: Finding the related parts of the file
Should we specify this in the spec? Garret: Worth mentioning but shouldn't be required
Will compare "naive" VCDIFF to "naive" Brotli and see if there are any advantages to the former. If not that will favor dropping VCDiff
Garret has been working with the Chrome team on a proper implementation
Garret will reach out about the current range-request spec status
Question: Are the compression formats in the HTTP spec optional, mandatory, or a mix?
(Brotli is in that list.)
Will plan on next meeting in two weeks
(February 7th)