See also: IRC log
<trackbot> Date: 09 June 2015
<scribe> scribenick: ChrisL
Vlad: we have more tables than
bits. do we swap out a less used one? or use a custom
table?
... we need to add meta
... we have OT tables, all known, then AAT which are somewaht
used, then odd stuff
ChrisL: the graphite tables are also there
kuettel: we have seen fonts with those tables
ChrisL: (explains Graphite)
RSheeter: easiest is to leave it alone and use 4 bytes
Vlad: we excluded vendor-specific tables used in production but not deployment
kuettel: when we did data gathering we only listed tables we actually saw
Vlad: ok so known table flags are full so any other table will use arbitrary tag flag plus actual tag, which is 4 bytes. no impact on functionality
sergeym: we know we use it a lot
and apple does too in their system fonts. no data on frequency
of use
... its okay
kuettel (shows list of observed table frequencies)
scribe: (looks for link)
s/fat fingers/agile, nimble fingers/
<kuettel> Here is the "Known Table Tags Proposal" thread from earlier
<kuettel> http://lists.w3.org/Archives/Public/public-webfonts-wg/2014Apr/0010.html
<kuettel> And here is the direct link to the spreadsheet with the underlying proposal data
<kuettel> https://docs.google.com/spreadsheets/d/111MT0l7LOVqotAnMXD4PMOm36jTPSznUigJPfxUYY_0/edit#gid=0
sergeym: meta is currently being defined
Vlad: next SC29 is in 2 weeks, could add to ammendment
sergeym: someone from Si Daniel's team is working on it
Vlad: (discussion of who on the
team is doing it)
... could you ask who is doing the proposal? end of this week
is the deadline
sergeym: we can send a mail to the list about it
Vlad: just need a draft spec language, and no objections
kuettel: in
https://docs.google.com/spreadsheets/d/111MT0l7LOVqotAnMXD4PMOm36jTPSznUigJPfxUYY_0/edit#gid=0
the grey ones wwre allocated early, the green ones are the new
OT tables
... color font ones from v.3
... after that, AAT and Graphite which we did find in use
... color font ones expected to be in use later
... so changes would affect fonts in the wild
Vlad: cvt is widely used
RSheeter: should not reallocate ones in the OT spec
kuettel: so everything allocated is used
resolved: leave known tables bitfield as it is
sergeym: ok
Vlad: even glyf table could be defined both ways, not forbidden. two parallel mechanisms to identify tables. no prefferential treatment
Vlad: aim is to have no undefined items at end of meeting
https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool
Vlad: get rid of the untestables and the no-longer a requirement ones
http://dev.w3.org/webfonts/WOFF2/spec/#conform-mustNotDuplicateTables
Vlad: it says physical tables, so if shared its in a collection directory, but physical table must only be presented once
sergeym: need to check our tools
do the right thing on all shared tables. if not shared on the
input font, do we want to force the woff2 encoder to fail the
check
... if it was shared in input font, it will be shared in
woff
... meaning smae data
... but then if there is the same data, that could have been
shared
RSheeter: don't want to do that
ChrisL: no, don't check for that
Vlad: each font can have it s own name table, but its not the tag but the offset that defines
ChrisL: don't want the wording to suggest a requirement for identity/duplicate table checking
(general agreement)
RSheeter: physical tables is a poor term
sergeym: if original font shared it, woff2 should share it
RSheeter: should not end up with more table offsets than you started with
Vlad: physical data must not be
duplicated, is the intent
... clear this up by removing the mention of input font
ChrisL: works for me
kuettel: this is the only mention of physical table, which is undefined
RSheeter: if tables in input have same offset, corresponding entries in the woff should point to the same index
(discussion of bizarre cases with unused duplicate tables that nothing points to)
(live spec editing)
<RSheeter> If two tables have the same offset in the input font, the corresponding indices in the CollectionFontEntry in the woff2 file MUST be the same.
(discussion on whether the validator should also look at the input sfnt as well as the output woff, and custom vs. general purpose validators)
ChrisL: edited https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustNotDuplicateTables
Vlad: make a dummy collection where all the entries are the same
RSheeter: if a table is physically duplicated and one copy unreferenced, it is a gap
(we need to check the gap requirement mentions collections)
<RSheeter> http://dev.w3.org/webfonts/WOFF2/spec/#conform-overlap-reject
(discussion on testing for extraneous data between tables in the compressed datastream)
Vlad: section 5 for compressed data format is the place for extraneous data mention
RSheeter: add a 5.5 specifiying restrictions on tables not having gaps
Vlad: or at the beginning section which applies to the whole datablock
kuettel: we need two, one FF and one AT
(live spec editing)
discussion of https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustTransformMultipeGlyfLoca
(editing https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustVerifyGlyfLocaShared )
(split into two tests)
https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustRejectSingleGlyfLocaShared seems okay
(or maybe not)
sergey we are discussing https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustRejectSingleGlyfLocaShared
(we split into two tests)
discussing http://dev.w3.org/webfonts/WOFF2/spec/#conform-mustRecordCollectionEntryIndex
https://www.w3.org/Fonts/WG/wiki/TestPlan20-AuthoringTool#mustRecordCollectionEntryIndex
RSheeter: in a number of places
in the spec we need to explicit about the array suffix
... such as starting at zero
<RSheeter> starting at zero is specifically for http://dev.w3.org/webfonts/WOFF2/spec/#conform-mustRecordCollectionEntryIndex (never explicitly said it started at 0)
<RSheeter> Array suffixes would be nice (at least for me) in http://dev.w3.org/webfonts/WOFF2/spec/#glyf_table_format, which tends to just use the name xxxStream to indicate it repeats, eg "UInt32 nContourStream" vs "UInt32 nCountourStream[nContourStreamSize]"
must use transform on glyph and loca seems to be missing from spec
<RSheeter> <span class="conform at" id="mustUseTransform"> needs conform- prefix in id
necessity to reference RFC6919 averted http://tools.ietf.org/html/rfc6919#section-4
<scribe> done with autoring tool test plan
https://www.w3.org/Fonts/WG/wiki/TestPlan20-Format#magicNumber
RSheeter: OMG it's full of assertions
<RSheeter> *gasp*
<scribe> Meeting: WebFonts Working Group f2f, Woburn
sergeym, we are going through https://www.w3.org/Fonts/WG/wiki/TestPlan20-Format
<scribe> ACTION: RSheeter to make a new github repo with the compiled output from the test generator [recorded in http://www.w3.org/2015/06/09-webfonts-minutes.html#action01]
<trackbot> Created ACTION-180 - Make a new github repo with the compiled output from the test generator [on Roderick Sheeter - due 2015-06-16].
Khaled, we are going through https://www.w3.org/Fonts/WG/wiki/TestPlan20-Format
editing it live
(discussion on desirability of a test case where the font data and the metadata are erroneously all compressed in one brotli stream)
Khaled: its not impossible to make this
kuettel: would the extraneous tests not catch this
Khaled: such a font should be rejected by the validator
https://www.w3.org/Fonts/WG/wiki/TestPlan20-UserAgent
(we discuss whether an unsorted table directory will cause a UA to not render a font, for excample because OTS rejects it)
table directory order and table order are changed from conformance requirements into a pointer to the OFF specification
ChrisL: we have conflated general
decoder tests and user agent (browser) tests. We should split
these into two conformance classes. Especially for TTC which
can be inspected and are supported in software, but nit
currently in browsers
... created https://www.w3.org/Fonts/WG/wiki/TestPlan20-Decoder
(discussion of extraneous-reject UA test)
(in particular that only inter-block extraneous data is covered in this section.)
https://www.w3.org/Fonts/WG/wiki/TestPlan20-UserAgent#mustRejectExtraData
(wondering what browsers do with a font with incorrect checksums)
<scribe> ACTION: vlad to add decoder category to woff2, and add to stylesheet [recorded in http://www.w3.org/2015/06/09-webfonts-minutes.html#action02]
<trackbot> Created ACTION-181 - Add decoder category to woff2, and add to stylesheet [on Vladimir Levantovsky - due 2015-06-16].
oh zakim, you lightweight
action-181?
<trackbot> action-181 -- Vladimir Levantovsky to Add decoder category to woff2, and add to stylesheet -- due 2015-06-16 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/181
I edited conform.css to add a fourth class, dc for decoder
close action-177
<trackbot> Closed action-177.
action-177?
<trackbot> action-177 -- Jonathan Kew to Propose two-bit per table version number -- due 2015-06-03 -- CLOSED
<trackbot> http://www.w3.org/Fonts/WG/track/actions/177
http://www.w3.org/Fonts/WG/track/actions/open
action-171?
<trackbot> action-171 -- Vladimir Levantovsky to Review conformance reqs to ensure they can actually be implemented -- due 2015-06-11 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/171
close action-171
<trackbot> Closed action-171.
action-818?
<trackbot> Sorry, but action-818 does not exist.
action-181?
<trackbot> action-181 -- Vladimir Levantovsky to Add decoder category to woff2, and add to stylesheet -- due 2015-06-16 -- OPEN
<trackbot> http://www.w3.org/Fonts/WG/track/actions/181
partly done (stylesheet part)
<scribe> ACTION: vlad to check with barbara re atypi-colocated meeting [recorded in http://www.w3.org/2015/06/09-webfonts-minutes.html#action03]
<trackbot> Created ACTION-182 - Check with barbara re atypi-colocated meeting [on Vladimir Levantovsky - due 2015-06-16].
(adjourned, agenda finished)
This is scribe.perl Revision: 1.140 of Date: 2014-11-06 18:16:30 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/vendore/vendor/ Succeeded: s/apople/apple/ FAILED: s/fat fingers/agile, nimble fingers/ Found ScribeNick: ChrisL Inferring Scribes: ChrisL Present: Vlad kuettel RSheeter sergey ChrisL Khaled Found Date: 09 Jun 2015 Guessing minutes URL: http://www.w3.org/2015/06/09-webfonts-minutes.html People with action items: rsheeter vlad[End of scribe.perl diagnostic output]