See also: IRC log
<scribe> scribe: ChrisLilley
kuettel: Rod can't get connected to the phone
Vlad: action-196?
action-196?
<trackbot> action-196 -- Vladimir Levantovsky to Review treatment of empty glyphs vs 0-contour glyphs -- due 2016-02-17 -- PENDINGREVIEW
<trackbot> http://www.w3.org/Fonts/WG/track/actions/196
Vlad: this will be an empty glyph on output
<jfkthame> (sorry to be so late)
Vlad: zero contour glyph (which
is very rare) with non-zero bbox, font is considered junk. It
is already a spwcial case, and has no paint. bbox should be
empty in that case
... safer to reject the font. We added a test case.
... for this one and for a missing one
... pass is no output on the encoder. Other test has a zero
bbox and produces an empty glyph on output
jfkthame: very much an edge case, sounds fine
(agreement)
close action-196
<trackbot> Closed action-196.
action-186?
<trackbot> action-186 -- Roderick Sheeter to Try time (decode) and size for null glyf/loca xform vs regular vs woff1 -- due 2016-03-02 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/186
RSheeter: maybe two weeks
action-172?
<trackbot> action-172 -- Chris Lilley to Register font media types -- due 2016-04-30 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/172
action-172 please chime in on the IETF list, new draft
https://github.com/svgeesus/ietf-justfont/issues
https://github.com/svgeesus/ietf-justfont
https://datatracker.ietf.org/doc/draft-ietf-justfont-toplevel/
kuettel: great to see this happening
jfkthame: about the fragment
syntax, css fonts
... putting the fragment in the css @font-face is good
(discussion on web architecture and where fragments are defined and how client-server works when there is a fragment)
action-195?
<trackbot> action-195 -- Roderick Sheeter to Check ua test https://www.w3.org/fonts/wg/wiki/testplan20-useragent#mustrejectinvalidbase128 -- due 2016-02-17 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/195
action-197?
<trackbot> action-197 -- Sergey Malkin to Investigate font collections; how are glyf/hmtx shared in practice -- due 2016-02-17 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/197
sergey: I am here
... have not found any fonts like that, still investigating.
Another week would be good.
Vlad: yesterday there was a suggestion for the new cts
RRSAgent: spec says not to trust those fields and we do in the current impl so we need tests for that
sergey: dedicated tests would be nice
RSheeter: spec says not to trust those fields and we do in the current impl so we need tests for that
<RSheeter> specifically WOFF2Header totalSfntSize and 'glyf's origLength
Vlad: construct a glyf table different size than original, can be smaller or larger. Will not give the exact number for target memory allocation size
RSheeter: maybe one too small, one too large
Vlad: not an error if you fail to optimise output of the decoder, or if you have a better optimisation
RSheeter: test should set the size to zero and it should still work
sergey: decoder rejects a perfectly valid font because of this issue.
Vlad: so it decodes to a valid
font
... need to define the conformance case for it
sergey: spec says these are only hints, is it enough?
Vlad: need to look at the text, make it more explicit. But just in plain English
ChrisLilley: think this one is easy enough to express as a conformance requirement
RSheeter: made a unit test easily
Vlad: its total sfnt size
RSheeter: and also orig length of
glyf
... Google code was trusting of that field because at the time
we hoped to have exact sizes
<RSheeter> (fix on it's way)
RSheeter: so can we have a conf test where orig length is set to an unfeasibly small value and check the font decodes correctly
Vlad: yes, just checking what the spec should say
sergey: use the same wording
jfkthame: can make test where those values are zero or one. Also huge values, and require the font is not huge full of empty space
RSheeter: agreed.
sergey: why, if it is valid (but huge)?
Vlad: any data entry in the table
entry can be doctored. can be done maliciously to try and do a
buffer overrun
... for example if it exceeds total sfnt size
sergey: prefer the test is that the font should not be rejected. don't care about the decoded size
RSheeter: yes
Vlad: I can do that, on the test
plan
... just a note on the total font size
sergey: this should be for any transformed table
Vlad: don't see a need for a cts here. Decoders will do much more complete tests to ensure they do not crash
sergey; we have valid fonts that fail because of sizes smaller than what the decoder produces
scribe: current code allocates that size. It fails.
Vlad: but that is an implementation bug
sergey: so therefore we want a test, and a stronger statement. must only be used for reference purposes. Must not reject the font.
Vlad: okay, we can do that. Change the note to a MUST NOT reject
<scribe> ACTION: vlad to add conformance clause for must not reject glyf size and original size [recorded in http://www.w3.org/2016/03/02-webfonts-minutes.html#action01]
<trackbot> Created ACTION-198 - Add conformance clause for must not reject glyf size and original size [on Vladimir Levantovsky - due 2016-03-09].
atypi and tpac coincide. atypi in warsaw, tpac in lisbon so possible to combine in one trip
Vlad: we have a choice, can
attend one or both so where do we have the meeting?
... tpac gives the option of a cross-group discussion
ChrisLilley: are there groups that want to talk to us?
Vlad: plan to attend both
ChrisLilley: so do I; easier if there is a meeting at atypi
(adjourned)
kuettel: there is a google office
in warsaw, if needed
... also we could takle post-woff2 plans at tpac
sergey: not sure i can meke it but will be online
jfkthame: do not usually attend
Vlad: tpac an excellent venue for
a what is next discussion with a wider group, developers,
anounce ahead of time to generate interest.
... perhaps one day of laundry and one of looking outside the
box
(adjourned for realz this time)
This is scribe.perl Revision: 1.144 of Date: 2015/11/17 08:39:34 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Found Scribe: ChrisLilley Inferring ScribeNick: ChrisLilley Present: ChrisLilley kuettel sergeym vlad rsheeter jonathan WARNING: Replacing previous Regrets list. (Old list: Ken_Lunde) Use 'Regrets+ ... ' if you meant to add people without replacing the list, such as: <dbooth> Regrets+ ken Regrets: ken Found Date: 02 Mar 2016 Guessing minutes URL: http://www.w3.org/2016/03/02-webfonts-minutes.html People with action items: vlad[End of scribe.perl diagnostic output]