| 08:54 | <annevk> | Hixie_: missed your modules question last night |
| 08:54 | <annevk> | Hixie_: http://wiki.ecmascript.org/doku.php?id=harmony:modules |
| 12:47 | <annevk> | Lets try this again. If we put Zip in the browser, what subset? Anyone know a good place to start asking this question? |
| 14:18 | <zewt> | annevk: as one limitation, i'd suggest random access mode only, not streaming--zip supports both, but they tend to want very different APIs |
| 14:23 | <zewt> | probably limit to store and deflate, ignoring legacy compressors (implode) and less widely supported ones |
| 14:27 | <annevk> | Requiring Zip64 seems like a good idea too. It probably depends on how this is going to be implemented. From scratch or using an existing library of sorts. |
| 14:27 | <zewt> | yeah i was going to say that |
| 14:27 | <zewt> | (zip64) |
| 14:27 | <zewt> | i don't think existing libraries matter |
| 14:27 | <zewt> | except for the deflate part |
| 14:28 | <zewt> | (which everyone has already anyway) |
| 14:30 | <zewt> | hmm, encoding for filenames is a tricky one |
| 14:32 | <zewt> | i don't think there's any standardization or anything in the file format whatsoever, which might mean having to say "must be utf-8, even though that dosen't match major clients" |
| 14:32 | <zewt> | (though i think that's what you happen-to-get with zip tools in *nix) |
| 14:32 | <zewt> | winrar assumes filenames are the local codepage, at least |
| 14:33 | <zewt> | 7-zip figures out utf-8 filenames in existing zips (probably heuristic) but encodes to the codepage |
| 14:34 | <zewt> | (heuristic since it seems to work with both utf-8 and codepage zips, in a quick test) |
| 14:36 | <zewt> | would somebody be writing a modern spec for the file format? the zip "appinfo" is actually really good for its time, but it doesn't define error handling, or end of central record searching |
| 14:37 | <Philip`> | http://www.w3.org/TR/widgets/#interoperability-considerations mentions some fun things about zips |
| 14:38 | <zewt> | oh there's an encoding flag? cool, wasn't aware of that |
| 14:38 | <zewt> | (not sure if it really helps, if other implementations don't use it) |
| 14:39 | <zewt> | i don't buy the "forbidden characters" thing, i think the whole "file names" section is bogus |
| 14:40 | <zewt> | if you have those characters then you can't extract them to native files in windows, that's all |
| 14:40 | <Ms2ger> | Oh, yes, unicode isn't exactly obvious with zips, I guess |
| 14:41 | Ms2ger | repressed that |
| 14:42 | <Ms2ger> | annevk, istr that omni.ja uses not-so-widely-supported features, might want to see what that all's about |
| 14:43 | <zewt> | crc checking is unobvious |
| 14:44 | <zewt> | it's tricky to fit into apis, since you only know if it's failed after you've read the whole file; maybe better to just explicitly ignore it |
| 14:45 | <zewt> | (otherwise an otherwise simple "blob = zip.getFile('foo')" API becomes weird) |
| 14:49 | <zewt> | annevk: are you thinking a ZIP API, or still trying to find a way to make zip URLs work? |
| 15:43 | <MikeSmith> | Ms2ger: I'm thinking of creating https://github.com/w3c/web-platform-tests/conformance-checkers for managing documents to test the validator |
| 15:44 | <MikeSmith> | good idea? bad idea? |
| 15:45 | <Ms2ger> | MikeSmith, doesn't it make more sense to put them under html/? (the ones for requirements in html, at least) |
| 15:46 | <MikeSmith> | yah but alternately makes sense to do conformance-checkers/html, conformance-checkers/svg etc. |
| 15:47 | <MikeSmith> | and that way we have all of them under one "conformance-checkers" root |
| 15:47 | <MikeSmith> | since they are different kinds of the tests than the browsers tests that are in the other dirs |
| 15:48 | <MikeSmith> | but I don't mind it either way |
| 15:49 | <Ms2ger> | I was thinking "what doesn't fit in this line: dom html IndexedDB navigation-timing conformance-checkers" |
| 15:49 | <Ms2ger> | But I guess that works too |
| 15:49 | <Ms2ger> | Maybe jgraham will suggest a filename convention |
| 15:50 | <MikeSmith> | yeah he does like his filename conventions |
| 15:51 | <MikeSmith> | I have one simple one already: foo.notvalid.html for tests that are intentionally invalid |
| 15:52 | <Ms2ger> | (I would suggest a manifest, of course) |
| 16:15 | <annevk> | zewt: format is relevant for both, and I think we want both |
| 16:17 | <annevk> | zewt: I'd like to avoid defining a format, but if we have to... |
| 16:46 | <gsnedders> | MikeSmith: +1 for a conformance-checkers root |
| 16:50 | <MikeSmith> | gsnedders: k |
| 16:51 | <gsnedders> | (Why? Because it keeps it out of the way for the more common use-case) |
| 16:53 | <MikeSmith> | right, that's part of what I was thinking too |