MLW-LT WG

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0108.html

HTML5+ITS validator

<fsasaki> scribe: fsasaki

jirka: used source code of the validator.nu
... currently best validator for html5
... validation engine is used by w3c too
... modified HTML5 to have ITS attributes in the validation preset
... will submit my code to validator.nu, so in the future validation with ITS will also be possible in W3C validator
... only issue: our working draft is not very stable yet
... we can postpone the addition to validator.nu until tpac meeting
... if you want to try yourself, I have posted links for offline-validaton

<declan> I think this is the github link: https://github.com/kosek/html5-its-tools

jirka: will be quite smooth to get our schemas part of w3c validator
... not so sure about validator.nu, but support by w3c validator will be sufficient
... another tool: created very simple tool that can take input HTML5 and produces XHTML
... that is, local ITS markup in XML
... useful if you have a toolchain that processes XML content
... not sure if it's very useful, haven't got a lot of feedback from implementors
... related question, also on agenda:
... currently our specification is missing a definition of local markup for XHTML
... for the past we used XML namespaces
... ITS markup was in ITS namespace
... in HTML5 we use its-* attributes
... so it makes more sense to use same prefixed attributes in XHTML
... I posted that issue on the ML
... outcome of discussion needs to be reflected in ITS 2.0 spec, that is: guidance how to use ITS local markup in XHTML

shaun: even with XHTML you would be able to use its-*
... instead of a namespace attribute
... XML syntax would still be OK
... which one is more useful may depend on the tool
... a tool that doesn't know about XHTML, you would want to transform those things into local markup
... using the HTML like markup might be more friendly to people doing HTML stuff

jirka: true, but we can do both approaches, but should have some guidance
... e.g. if you use XHTML internally, in a workflow, you can use the ITS namespace
... if it is a public web site, prefixed attributes are better
... in this way you have consistent parsing even with HTML parser
... so both approaches are OK, but we would need some explanation in the spec
... there is "polyglott markup", that is: intersection of XHTML and HTML syntax
... if you use that syntax, you get the same parsing result with both HTML and XHTML parser
... if you want to use a document in this subset, you need to use its-* attributes
... ITS namespaced attributes are not allowed in polyglot
... I don't have a preference, but we should mention this in the spec

shaun: sounds good

pedro: for us it is very important to have an option for XHTML

<scribe> scribe: jirka

<scribe> ACTION: Jirka to draft section about using its-* and its: in XHTML [recorded in http://www.w3.org/2012/09/20-mlw-lt-minutes.html#action01]

<trackbot> Created ACTION-220 - Draft section about using its-* and its: in XHTML [on Jirka Kosek - due 2012-09-27].

felix: schema refactoring will wait after prague meeting, so spec is more stable

Extensions of HTML5

<fsasaki> scribe: fsasaki

jirka: one problems of HTML5 is lack of extensibility
... in XML you can easily use namespaces
... in spring we had decide which approach to use in HTML5
... we decided for its-* attributes
... there was some support for this approach in HTML5 spec
... I wasn't 100% sure that if we use such attributes it will be well received by HTML community
... there are not so many specs that add something to HTML5
... yesterday chairs of HTML WG introduced a new plan
... goal is to speed up HTML5 recommendation
... one approach is: split some problematic and other parts into extension specifications
... at the same time they say that the validator will support such extensions
... so validator (= w3c validator) will support also elements and attributes from widely supported extensions
... that is good for us because it is common to have such extensions
... also, mike smith from w3c, said to felix that he has nothing against adding support HTML5+ITS2.0 once we have schemas ready
... I'd just like to postpone this until all data categorie are settled
... so for us it is good
... there was some discussion on HTML mailing list
... saying extensions are normal for HTML5, and having a spec HTML5+ITS is OK thing
... and validator can support that out of the box

<scribe> scribe: jirka

felix: we still have plan to meet with HTML WG at TPAC
... Felix is in touch with HTML WG chairs on date/time of joint meeting

Pedro: We are trying to move clients to HTML5 but we want to prevent compat. problems

Felix: Now you can use either XHTML or HTML with ITS, depending on what's more approapriate for your client

PRAGUE PREPARATION

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0086.html

Felix: Dave developed test suite wiki page

<fsasaki> http://www.w3.org/International/multilingualweb/lt/wiki/Test_Suite_Development

<fsasaki> http://phaedrus.scss.tcd.ie/its2.0/its-testsuite.html

Felix: we need to get implementors in Prague in order to fill above table
... feedback from W3C about test suite design - it's OK
... only tests shouldn't be so artificial

<fsasaki> http://phaedrus.scss.tcd.ie/its2.0/inputdata/translate/html/translate2html.html

<fsasaki> http://phaedrus.scss.tcd.ie/its2.0/expected/translate/html/translate2HtmlOutput.txt

Felix: we need at least two outputs for each data category

Declan: Output is not useful itself, it's just sort of metadata?

Felix: Yes, we changed this so it is easy to compare output from various tools.
... you should do also test of real life usage

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0118.html

Felix: above is template for gathering basic information about implementations

Overriding semantics

see thread http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0092.html

Yves: There is no formal definition of overriding in ITS 1.0/2.0. But there is something in tests for ITS 1.0.

<fsasaki> comment from felix: tests = ITS 1.0 tests

Yves: so intent is here in tests for ITS 1.0, we should be backward compatible with ITS 1.0

Olaf: Overriding non-existing value seems quite complicated. I prefer to override only existing values.

Yves: Yes, this would be more intuitive, but it is not compatible with ITS 1.0.

Olaf: Can we change this? It seems that noone implementd this anyway.

Yves: Overriding can get very complex if we start to use pointers and references as well.

Olaf: There must be way how to fix past mistakes.

Felix: We need to discuss this in Prague.
... in the scope of affected categories

Tadej: Example - one document annotated by two tools.
... is this legal, can both tools annotate the same element/attribute?
... we can allow only global annotations

Olaf: We can consider additional global parameter...
... (sorry I haven't catched it)

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0101.html

Declan: agrees with Olaf

<fsasaki> look at the lq-issue in above mail

felix: describes problem - is inherited global value of data category if only part of it was locally overriden

Olaf: If you set set of parameters globally, you should be able to override part of it only

Felix: We can have document level annotations

<fsasaki> Jirka thinks that we need terminology here

Declan: that would work for mtConfidence

<fsasaki> Jirka we don't have other parameters then its:param in ITS

<fsasaki> Jirka What about compound data category

Tadej: that would work for me as well

<scribe> ACTION: felix to Summarize discussion about overriding [recorded in http://www.w3.org/2012/09/20-mlw-lt-minutes.html#action02]

<trackbot> Created ACTION-221 - Summarize discussion about overriding [on Felix Sasaki - due 2012-09-27].

Meeting adjourned