15:11:39 <RRSAgent> RRSAgent has joined #webfonts
15:11:39 <RRSAgent> logging to http://www.w3.org/2014/10/28-webfonts-irc
15:11:41 <trackbot> RRSAgent, make logs world
15:11:41 <Zakim> Zakim has joined #webfonts
15:11:43 <trackbot> Zakim, this will be 3668
15:11:43 <Zakim> I do not see a conference matching that name scheduled within the next hour, trackbot
15:11:44 <trackbot> Meeting: WebFonts Working Group Teleconference
15:11:44 <trackbot> Date: 28 October 2014
15:12:05 <ChrisL> zakim, remind me in 4 hours to go home
15:12:05 <Zakim> ok, ChrisL
15:12:13 <ChrisL> rrsagent, here
15:12:13 <RRSAgent> See http://www.w3.org/2014/10/28-webfonts-irc#T15-12-13
15:12:42 <ChrisL> chair: Vlad
15:12:59 <ChrisL> Meeting: WebFonts f2f meeting, California
15:13:52 <ChrisL> scribenick: ChrisL
15:14:04 <ChrisL> topic: compression
15:15:23 <ChrisL> zoltan: excited to see woff progress. http compression is helped by a command line tool separate from font use. compiled and ready to go
15:15:50 <ChrisL> ... ideas to improve the faster codec
15:16:17 <ChrisL> ... slower one is the open source one, has better compression
15:16:40 <ChrisL> vlad: will the faster one be shared also?
15:16:44 <ChrisL> zoltan: yes
15:16:58 <ChrisL> vlad: compression is fast but slow, people complain
15:17:17 <ChrisL> zoltan: faster codec uses a lot of ram
15:17:48 <ChrisL> raph: same wire format?
15:18:03 <ChrisL> zoltan: yes, but does not use all compression features
15:18:29 <ChrisL> raph; any spec changes for this?
15:18:33 <ChrisL> zoltan: no
15:19:06 <ChrisL> vlad: I have a gift. its a big font. please compress, it crashes on compression
15:20:39 <ChrisL> ChrisL: good to get the brotli results for monotype, adobe, and the two fond providers
15:21:05 <ChrisL> zoltan: paged mark adler and asked if he would review it, no response yet
15:21:55 <ChrisL> ... he reviewed webp lossless, is a great reviewer
15:23:00 <ChrisL> ChrisL: asked j-l gailly?
15:23:07 <ChrisL> zoltan: yes, no response
15:23:32 <ChrisL> ... he is not working on compression since 2006
15:23:49 <ChrisL> ... mark adler is still actively involved
15:25:45 <ChrisL> kuettel: show the new top-level brotli site
15:26:22 <raph> https://github.com/google/brotli
15:26:48 <raph> (soft launch for now)
15:27:17 <Vlad> Vlad has joined #webfonts
15:28:21 <ChrisL> (discussion on open sourcing the fast codec as well)
15:29:19 <kuettel> Kenji found the following JavaScript wrapper implementation via Google Alerts:  https://www.npmjs.org/package/brotli
15:29:22 <ChrisL> kbx: there is a javascript one
15:29:38 <Vlad> s/fast but slow/good but slow/
15:29:43 <ChrisL> (discussion on whether this is a wrapper, a port or a new impl)
15:30:22 <ChrisL> links to http://www.ietf.org/id/draft-alakuijala-brotli-02.txt which is the third version
15:33:54 <ChrisL> behedad: want to make sure the chrome deployed code matches the repro
15:34:26 <ChrisL> Vlad: binary mode should not be used for metatata, myfonts is adding woff metadata
15:34:46 <ChrisL> zoltan: decompression is the same either way
15:35:45 <ChrisL> raph: tools should be updated to use the best mode
15:36:16 <ChrisL> behedad: any plan to adress compression startup time?
15:36:31 <ChrisL> zoltan: not planned currently but we could do
15:36:55 <ChrisL> kbx: open bug on chrome to switch to new github brotli, do it now?
15:37:03 <ChrisL> zoltan: yes
15:37:19 <sergeym> sergeym has joined #webfonts
15:37:41 <ChrisL> behedad: chrome uses the copy in ots
15:37:59 <ChrisL> ... which gets used in firefox and chrome
15:38:56 <ChrisL> behedad: what is the testing plan to avoid regressions over time
15:39:16 <ChrisL> kuettel: current reference is good at doing round trip sanity checks
15:39:47 <ChrisL> behedad: hard to see if patches break anything
15:40:16 <ChrisL> (discussion on open sourcing the brotli test data)
15:41:12 <ChrisL> kuettel: there is a new project for woff2 which is unpopulated yet
15:41:31 <ChrisL> behedad: reference should be updated and ots updates from that
15:41:41 <ChrisL> raph: and right now there is some divergence
15:42:06 <ChrisL> behedad: firefox uses the github fork, chrome has not updated yet
15:42:39 <ChrisL> zoltan: mostly encoder changes, minor improvements
15:44:01 <ChrisL> behedad: brotli we need to get to users for adoption
15:44:24 <ChrisL> Vlad: best for brotly to have the fast compression released asap otherwise they think its always slow
15:44:52 <ChrisL> behedad: startup takes two seconds, for the text mode
15:46:28 <ChrisL> raph: static content can use the slower compression for best compression
15:46:43 <ChrisL> ... startup hit is for text mode only
15:47:04 <ChrisL> zoltan: can remove dict transforms to speed it up
15:47:17 <ChrisL> ... 100 times faster
15:47:27 <ChrisL> ... loose 0.5% on compression
15:48:23 <ChrisL> Vlad: for dynamic compression of a font subset, that balances some of the compression loss
15:48:52 <ChrisL> kbx: is there an advantage to using text encoding?
15:49:00 <ChrisL> .. for html, css, javascript
15:49:17 <ChrisL> zoltan: much better for a few k of text
15:49:42 <ChrisL> Vlad: between text mode and binary mode brotli?
15:50:09 <ChrisL> zoltan: uses utf-8, using static disctionary or not
15:50:39 <ChrisL> ... static very goof for vocabs like css, html etc
15:51:10 <ChrisL> ... can give 30% in binary size
15:51:16 <ChrisL> s/goof/good
15:52:44 <ChrisL> zoltan: decompress is faster if a dict was used. so slow encode gives fast decode
15:52:53 <ChrisL> ... memcpy of dict entries
15:54:42 <ChrisL> Vlad: at some point can brotli be an ISO standard via MPEG, lots of compression experts
15:55:57 <ChrisL> zoltan: could do but dont want to spend a lot of time there
15:56:46 <ChrisL> Vlad: ground is prepared, takes about 18 months elapsed, process wise
16:01:23 <ChrisL> behedad: not clear what the value of ISO standardization is
16:01:51 <ChrisL> Vlad: iptv for example uses only ISO standards
16:03:25 <ChrisL> (discussion on value of iso standards)
16:03:47 <ChrisL> (chris updates woff spec to point to brotli -02)
16:05:37 <raph> (shouldn't you point to http://www.ietf.org/id/draft-alakuijala-brotli which is always the latest draft?)
16:06:30 <ChrisL> (reminiscences on PFR)
16:06:45 <ChrisL> (hmm maybe)
16:06:56 <ChrisL> (except the reference needs a date)
16:07:43 <raph> (ah)
16:09:59 <ChrisL> topic: nominal size
16:10:43 <ChrisL> raph: size of reconstructed tt data is tricky as many decompressions with same semantic meaning on point data
16:10:55 <ChrisL> ... encoding of flags bytes, alighnment and padding
16:11:15 <ChrisL> Vlad: and coordinate point representation
16:11:25 <ChrisL> raph: more about the flags than the coordinates
16:12:20 <ChrisL> behedad: we already aboid unecessary padding
16:12:55 <ChrisL> raph: goal on nominal size is to allocate a chunk of memory and know how much is required for an unsophisticarted decoder and not reallocate
16:13:11 <ChrisL> ... and also to decide not to decompress
16:13:42 <ChrisL> ChrisL: no guarantee its big enough
16:14:00 <ChrisL> Vlad: incoming size is known, efficiency is not known
16:14:13 <ChrisL> ... nominal size amy be enough but is not a guarantee
16:14:20 <ChrisL> raph: can be a guartantee
16:15:11 <ChrisL> raph: its modelling a simple but not brain-dead decoer. always most compact point rept, greedy flags
16:15:33 <ChrisL> ... recommended alignment (in OT is obselete)
16:16:43 <ChrisL> raph: ultimate gosl is a rel impl will get inside the nominal size without much work
16:16:53 <ChrisL> its therefor e anominal not a minimal size
16:17:44 <ChrisL> Vlad: create an open ended environment, impl makes its own choices on how efficient to be
16:18:06 <ChrisL> ... security aspect of review means buffer overrun is an issue
16:18:31 <ChrisL> ... conern was a false sense of security about overrun
16:20:13 <ChrisL> raph: level of guarantee: nominal size is deterministic. algo produces the nominal size. its not the original size.
16:20:29 <ChrisL> ... and that size is easy to get smaller than
16:20:55 <ChrisL> ... impl can reallocate, or can ignore and do dynamically. so no harsh requiremment
16:21:24 <ChrisL> raph: can alloc to nominal size and will use some size less than that
16:21:50 <ChrisL> ... if so its guaranteed not to get larger
16:22:05 <ChrisL> ... if you do, and exceed it, the font can be rejected as invalid
16:23:02 <ChrisL> ChrisL: the invalid if over makes me a lot more comfortable
16:23:30 <ChrisL> behedad: can argue problem outweighs benefits
16:23:56 <ChrisL> Vlad: if you can modify the text to make it clear that it helps implementors
16:24:10 <ChrisL> ... that bit is informational
16:24:28 <ChrisL> ... then define the nominal size and give the algorithm and has no smarts
16:24:37 <ChrisL> so if you ever go over that the font is invalid
16:25:50 <ChrisL> raph: ok that makes sense
16:26:15 <ChrisL> raph: first impl just used the original size, as it is known
16:26:33 <ChrisL> .. decoder used 4byte alignment, and so we tended to run over and fsail
16:26:45 <ChrisL> ... not so much on flags, but the alignment
16:26:55 <ChrisL> ... many fonts have tighter alignment
16:27:20 <ChrisL> ... chrome decoder switched to 2 byte alighnment.
16:27:58 <ChrisL> ... (long and short loca tables) hyper omptimizer may choose a 1 byte alighnment omn larger tables
16:28:37 <ChrisL> raph: so around november 2013 there was a normalizer which computed the nominal sise in the encoder
16:28:57 <ChrisL> ... except currently specced to 4 byte alignment, should be 2
16:30:10 <ChrisL> ... released encoder computes true nominal size, so guaranteed not to reject
16:30:40 <ChrisL> raph: deployed decoders use a 2 byte alignment
16:31:23 <ChrisL> "original" length is actually nominal length
16:33:43 <ChrisL> raph: original length should be discarded, and encode only nominal length
16:33:55 <ChrisL> Vlad: only if its a 100% giuarantee
16:34:00 <ChrisL> raph: it is
16:34:21 <ChrisL> RSheeter: so we add MUST reject if larger
16:34:24 <ChrisL> (yes)
16:34:57 <ChrisL> ACTION raph to work with vlad to clarify all uses of original length
16:34:58 <trackbot> Created ACTION-153 - Work with vlad to clarify all uses of original length [on Raph Levien - due 2014-11-04].
16:36:17 <ChrisL> raph: better to rename rather than clarify. so change it to decoded length
16:37:20 <ChrisL> behedad: impl may have to go from 2byte to 4byte and then will need to overide that value if its 130k or larger
16:37:40 <ChrisL> raph: in that case need to rewrite local table
16:38:14 <ChrisL> behedad: should ignore what head dtable says depending on what you are decoding to
16:38:36 <ChrisL> ... set to ewhat they are using in the decoder
16:39:22 <ChrisL> Vlad: if you are right on the boundary)
16:39:34 <ChrisL> behedad: must not reject the font in tht case
16:40:05 <ChrisL> raph: agree this is what we should do
16:40:24 <ChrisL> or make the loca table agree with the nominal
16:40:34 <ChrisL> Vlad: its decoder specific
16:41:02 <ChrisL> raph: do want tocheck chrome impl decodes when the loca table format is wrong at the 130k boundary
16:42:56 <ChrisL> Vlad: proper for spec t glag that
16:43:03 <ChrisL> raph: easy for an impl to miss
16:47:07 <ChrisL> topic: ttc
16:47:23 <ChrisL> RSheeter: seemed to work, but should we or not
16:49:20 <ChrisL> topic: testing
16:49:27 <ChrisL> (explains how woff1 worked)
16:50:52 <ChrisL> action: chris to make a w3c woff2 test repo
16:50:52 <trackbot> Created ACTION-154 - Make a w3c woff2 test repo [on Chris Lilley - due 2014-11-04].
16:50:58 <ChrisL> topic: ttc
16:51:27 <ChrisL> raph: no language for ttc is in the current spec. nice to have, covers all of OT standard
16:51:47 <ChrisL> ... no browsers do ttc
16:52:12 <ChrisL> Vlad: concluded earlier its nice to have, forn collections not ttc
16:52:18 <ChrisL> raph: actually OTC
16:52:35 <ChrisL> Vlad: iso v3 now just calls them collections
16:52:50 <ChrisL> ... in woff1 were not supported sand noone cared
16:53:25 <ChrisL> Vlad: expect collections become more widely used, so should not be impossible
16:53:39 <ChrisL> raph: expands woff2 over woff1
16:53:47 <ChrisL> Vlad: any wire format changes?
16:53:49 <ChrisL> raph: yes
16:53:58 <ChrisL> RSheeter: same for things that are not collections
16:54:33 <ChrisL> RSheeter: flavor in woff2 header, look for collection
16:54:51 <ChrisL> (we check about flavor)
16:55:02 <ChrisL> raph: ttc flavor is not normative
16:55:49 <ChrisL> Vlad: flavor is sfnt version
16:56:10 <ChrisL> behedad: first 4 bytes defines collection
16:56:36 <ChrisL> raph: sfnt version is not as normative way to distinguish ttc
16:56:59 <ChrisL> .. can use for all fonts we know. ttc tag does not go into woff2 header
16:58:27 <ChrisL> RSheeter: flavor of the parts of the collection ges elesewhere
16:58:41 <ChrisL> raph: extended table forma t with duplicated blocks
16:59:11 <ChrisL> RSheeter: not clear if only one of loca and glyf is shared
16:59:32 <ChrisL> behedad: what does otc use as a tag?
16:59:58 <ChrisL> RSheeter: some smaller sample files would help
17:02:05 <ChrisL> ChrisL: we are ahead of the browsers here. there is no reqt to support ttc in browsers
17:02:25 <ChrisL> Vlad: ttcf is iused for ot collections as well
17:03:17 <ChrisL> behedad: is it right to do this now, its less mature. argued for this in woff1. why not do it separately
17:04:13 <ChrisL> kbx: how do you use it in css
17:04:20 <ChrisL> (a fragment identifier)
17:06:44 <ChrisL> (discussion on ttc use cases, mostly on han)
17:07:25 <ChrisL> s/han/ideographs/
17:09:04 <ChrisL> Vlad: collections not much in use but increasing, eg for traditional and simplified chinese
17:09:20 <ChrisL> Vlad: could recharter to add as a new work item.
17:10:15 <ChrisL> (rechartering risk analysis)
17:14:57 <ChrisL> Vlad: we can reserve on flavor
17:16:21 <ChrisL> kuetel: agree with ChrisL its better to do now from a marketing/adoption perspective
17:20:59 <ChrisL> resolved: we will support collections in WOFF 2.0
17:21:20 <ChrisL> raph: will do code review on RSheeter prototype
17:21:26 <ChrisL> ... about a page of spec added
17:21:46 <ChrisL> ... one para defining what to do on flavor
17:22:21 <ChrisL> raph: there is aproposal in email, uses indices not byte offsets so no overlap
17:22:45 <ChrisL> behedad: can you load one face without loading the other faces
17:22:56 <RSheeter> http://lists.w3.org/Archives/Public/public-webfonts-wg/2014Feb/0018.html
17:23:17 <ChrisL> action-154?
17:23:17 <trackbot> action-154 -- Chris Lilley to Make a w3c woff2 test repo -- due 2014-11-04 -- OPEN
17:23:17 <trackbot> http://www.w3.org/Fonts/WG/track/actions/154
17:23:20 <kuettel> kuettel has joined #webfonts
17:23:21 <ChrisL> https://github.com/w3c/woff2-tests
17:23:29 <ChrisL> close action-154
17:23:29 <trackbot> Closed action-154.
17:24:55 <ChrisL> kuettel: brotli decompress is all one, but after than you can select
17:25:11 <ChrisL> raph: 90% is common anyway so no gain there
17:26:32 <ChrisL> resolved: even for collections, one brotli stream for the whole collection
17:27:01 <ChrisL> rrsagent, make minutes
17:27:01 <RRSAgent> I have made the request to generate http://www.w3.org/2014/10/28-webfonts-minutes.html ChrisL
17:33:07 <ChrisL> topic: recap
17:33:31 <ChrisL> raph: should nit require  adecoder to reject the font if its over nominal size
17:34:00 <ChrisL> ChrisL: thought the main issue as a malicious font trying to overrun memory
17:34:05 <ChrisL> raph: that is a concern
17:34:31 <ChrisL> Vlad: we decoided its a guaranteed largest size. easier to see when we have a firm spec
17:34:59 <ChrisL> behedad: may be abale to see how far from ideal the nominal is
17:35:29 <ChrisL> raph: reluctant, might want to aligh on wierd boundaries
17:37:07 <ChrisL> Vlad: lets start with the spec part and then decide on rejecting oversized fonts
17:37:51 <ChrisL> raph: decoder MAY reject the font
17:39:42 <kbx_> kbx_ has joined #webfonts
17:42:19 <RSheeter> github is rsheeter
17:43:22 <RSheeter> (would also want khaledhosny, or permission to add addtl users [preferred])
17:47:52 <ChrisL> rrsagent, make minutes
17:47:52 <RRSAgent> I have made the request to generate http://www.w3.org/2014/10/28-webfonts-minutes.html ChrisL
17:49:20 <ChrisL> Vlad: so in short term not going to last call.
17:49:31 <ChrisL> ... for cts, draft test descriptions are there
17:49:55 <ChrisL> (adjourned)
17:50:00 <ChrisL> rrsagent, make minutes
17:50:00 <RRSAgent> I have made the request to generate http://www.w3.org/2014/10/28-webfonts-minutes.html ChrisL
18:35:08 <ChrisL> ChrisL has joined #webfonts
18:58:14 <kbx> kbx has joined #webfonts
19:12:06 <Zakim> ChrisL, you asked to be reminded at this time to go home
19:33:23 <Zakim> Zakim has left #webfonts
20:06:49 <jdaggett> jdaggett has joined #webfonts
20:08:13 <jdaggett> jdaggett has joined #webfonts