<Vlad> https://www.w3.org/Fonts/WG/track/actions/open
<Vlad> scribenick: Persa_Zula
Vlad: hoping to get an update
from all; and timeline of future development ideas
... got updates from Myles and he's feeling better. Is asking
for someone to help with his efforts
Garret: Can help
Vlad: The tool that needs
completed is to pre-process fonts for optimizing fonts. Quality
of the output of that tool will determine the performance of
our solution
... if that critical component is developed by Myles, we can
make sure there might not be issues that are based on who
builds the tool
Garret: can hook the tool up with
the simulation codebase and when he makes improvements on the
tool it can determine how the code changes affect the
performance
... will connect with Myles
Vlad: let's use the email list
Garret: got permission to share results based on the data analysis. Will have resultset ready for the next meeting
Vlad: my action item - trying to
get folks on his team to help with OEM fonts in their online
database. If they can't figure this out, he'll have to use a
different method to get the fonts to the WG. Some of these are
composite pan-CJK fonts, and the test font from Kamal which was
shared last time
... Might take another week to sort this out with his team
Garret: our final results aren't done yet; results are very heavily weighted towards latin. Would like to break it out per language to get better results
Vlad: Have you tried running simulation with 1 variation through the different approaches?
Garret: just ran using unicode codepoints as the primary driver; haven't tried the codepoints to glyphs hybrid approach
Vlad: more excited about this
approach now; watching email exchanges from other WG and
privacy concerns keep coming up. TTML WG is publishing an
update and the Candidate recommendation; possibly using fonts
as a fingerprinting tool? only bringing it up because people
bring up privacy more and more.
... requesting charsets for an individual font could be seen as
a privacy violation; some languages; every time you ask for a
character subset you can guess what the text you're looking at
could be saying
... glyph ID requests are free of this issue
Garret: glyph IDs might not be
enough because the source font could be retrieved and you could
still get back to the mapping
... blocking up the lower use codepoints into groups might be a
lot of help here. Hasn't implemented this yet but might want to
look into it again
<chris> Agree that reverse mapping glyphid to character code is tractable, just a bit more work
Vlad: there are more approaches we can look at, but there might be small performance hits to implement them. Just something we need to keep in mind. By the time we get to the recommendation that this will go through the review process and privacy will be brought up
jpamental: agree that the approach that we can request more than necessary
Garret: Agreed that we need to cover this as we move forward
Vlad: let's revitalize the email list and get more things done on the mailing list
Garret: one last quick thing -- playing around with desubroutinized or subroutinized fonts -- desubroutinized fonts end up with bettter performance at the end of the day
Vlad: is this because of compression or an extra layer of complexity to go through?
Garret: not sure why this is happening yet, but the guess is not having the extra layer of complexity might be a win
Vlad: with woff2, desubroutinzed compressed better
Garret: since we're using brotoli, this might be the case here too
Vlad: any working from home anecdotes?
Garret: all is well here, not missing the office too much :)
Vlad: initial optimistic view of
short lived situation has changed, went back and raided his
office and brought the stuff home :)
... if we want, we can do future video conference calls
... let's use email list to help us get where we want to be
sooner
... will ping Myles and see when we can get him in the meeting
again, we'll get the details of the next meeting on the mailing
list
... tenatively 2 weeks from now, maybe next week
This is scribe.perl Revision: 1.154 of Date: 2018/09/25 16:35:56 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) Present: Vlad chris Persa_Zula Garret jpamental Found ScribeNick: Persa_Zula Inferring Scribes: Persa_Zula Found Date: 30 Mar 2020 People with action items: WARNING: IRC log location not specified! (You can ignore this warning if you do not want the generated minutes to contain a link to the original IRC log.)[End of scribe.perl diagnostic output]