MLW-LT WG -- 04 Mar 2013

High level usage scenarios publication

XLIFF+ITS namespace discussion, other mapping topics

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Mar/0034.html

fsasaki: Felix had action item to check whether W3C can host the namespace; alternative was to use the itsx namespace
... David mention that itsx was used by other applications, so it would need to be a different namespace

daveL: we encountered similar things in provenance, ITS ontology;
... having the extension in a separately addressable file made sense for us

<fsasaki> http://itstool.org/extensions/

fsasaki: itst is currently used by the ITS tools,
... People should be able to create their own extension of the namespace, with the ability to have stand-alone files for the namespaces

daveL: can we have sub-spaces there? /rdf, etc?

fsasaki: yes, since it's an URI

<daveL> http://www.w3.org/International/multilingualweb/lt/wiki/Provenance_Best_Practice#Extraction

daveL: here's an example of how we used extensions in our provenance RDF example
... we want to be able to use it both in RDF use cases, as well as XLIFF

fsasaki: Why would the namespace for the RDF be different as for XLIFF?

daveL: we could manage them separately, not sure if any other difference
... in the RDF one, we have 'domain', but it wouldn't map to an existing ITS XML attribute

fsasaki: would you make use of 'domain' data type when validating RDF?

daveL: yes, you could actually do this within RDF

fsasaki: RDF tools usually don't resolve the XSD to enforce data types
... would people like an XSD on the namespace URI location?

daveL: David should know more on that

Locale filter exclusion

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Feb/0320.html

<fsasaki> <p its-locale-filter-list="*-CA">Legal notice for Canada</p>

fsasaki: we want an ability for some content to be excluded; yves proposed using the ! operator for 'excluding' a particular locale

<fsasaki> <p its-locale-filter-list="!*-CA">Legal notice for all locales except Canada</p>

shaunm: we had that functionality, but no-one wanted it

fsasaki: we dropped it in August, and we didn't have a clear record of discussion beyond private e-mails

Yves_: i hadn't followed up on that topic
... but it is still useful functionality

fsasaki: would that result into additional changes in the test suite?

Yves_: current test cases only include locales, we'd need to add a new test case for excluding

shaunm: it's a fairly big change for a last call

fsasaki: we shouldn't avoid doing that, we don't have to make a final decision today

Des: I support the ! exclusion proposal

fsasaki: we have 5 implementers for locale filter

<omstefanov> an excluding test case should show excluding of more than 1 case, such would be realistic. The example Yves brought is so "real world" it's hard to believe we missed it.

fsasaki: do people have a preference for the ! over a special attribute?

<omstefanov> what would the attribute look like, and would it be IN ADDITION TO positive selections?

shaunm: attribute also isn't problematic

<Ankit> either way (! or keyword) wors for us (DCU)

Yves_: personally think ! is easier, but not against a specific attribute

<clieske> BCP47 does not provide an impression for "inverse"?

<fsasaki> no

<clieske> Should it? :-)

clieske: this is a suggestion for providing feedback to BCP47 people for including an exclusion syntax

<scribe> ACTION: felix to check with I18N group about exclamation mark to signify an exclusion [recorded in http://www.w3.org/2013/03/04-mlw-lt-minutes.html#action01]

<trackbot> Created ACTION-454 - Check with I18N group about exclamation mark to signify an exclusion [on Felix Sasaki - due 2013-03-11].

<scribe> ACTION: Yves to create examples of locale filters with exclusion (using exclamation marks) [recorded in http://www.w3.org/2013/03/04-mlw-lt-minutes.html#action03]

<trackbot> Created ACTION-455 - Create examples of locale filters with exclusion (using exclamation marks) [on Yves Savourel - due 2013-03-11].

HTML and ITS discussion

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Feb/0298.html

Yves_: some html elements have counterintuitive behaviour on inheritance, for instance <del> shouldn't be translatable, while its sub-elements can have different defaults
... we shouldn't make things like this compicated for the user

kfritsche: it makes sense to implement this as defaults, not as defaults

shaunm: we could make the same argument for every format
... it becomes a problem on how we apply information

fsasaki: in practice, many tools have HTML filters, which describe defaults different to the defaults in the format-agnostic ITS spec
... for instance, the translate in HTML will have different defaults than its:translate

<fsasaki> https://www.w3.org/Bugs/Public/show_bug.cgi?id=21085

<fsasaki> above is the HTML5 bug

Yves_: shaun has a point, but in this case we write against the HTML5 spec, which we refer to specifically. it might make sense to have different defaults for that particular format

fsasaki: another argument might be the workflows of Linguaserve, where XHTML is used in the XML processing chain, as well as HTML processing chains, and having them aligned can be beneficial
... the question is also for the xml-only implementers: will they be required to implement these defaults

Pedro: yes, we have this use case

fsasaki: we need some more feedback on this on mailing list. let's keep the topic on the agenda and see how the discussion moves forward

Testing - good news

<fsasaki> http://htmlpreview.github.com/?https://raw.github.com/finnle/ITS-2.0-Testsuite/master/its2.0/testSuiteDashboard.html#conformance-classes-overview

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt-tests/2013Mar/0000.html

<fsasaki> http://www.w3.org/2012/09/mlw-lt-charter.html

fsasaki: here's the test suite status

fsasaki: our charter says that we do testing from March to October, and we're mostly done. We can take time to finalize some features, and I'm confident we won't need much time for testing

philr: one topic from last week was discussing which way should the stand-off references point?

fsasaki: I made the experiment from inverting the pointer direction, from outside to inside might be more efficient

<scribe> ACTION: phil to follow up on stand-off pointer direction (quality, provenance) [recorded in http://www.w3.org/2013/03/04-mlw-lt-minutes.html#action05]

<trackbot> Created ACTION-456 - Follow up on stand-off pointer direction (quality, provenance) [on Phil Ritchie - due 2013-03-11].

Test suite errors, please fix if you see your name

fsasaki: some errors aren't really errors, for instance RDFa attributes
... the output will likely influence the output of the test suite
... so having an update in time will be helpful

clarify character encoding behaviour see latest thread

swalter: the discussion split in several threads. we made some specification on how a program should behave when not supporting an encoding given constraints, there wasn't any consensus on what should be done with the storageSize constraint
... it usually denotes the size of the database field, and it's not a MUST. how can we have a MUST, if it's not required behavior

fsasaki: if we feel that a MUST is not necessary, we can reject the comment

shaunm: what should the program do if it doesn't understand the character encoding?

swalter: the program uses the encoding to verify the constraint, but it's not obligatory to validate that
... As an user of ITS, you could feel unsafe whether you give a tool a constraint which it will not obey
... especially when there are consumers that may care
... however, we normally don't constrain consumers

Yves_: we could have a note that would describe the expected behaviour (described with lowercase should, must)
... and not worry about conformance at our level

swalter: there could be a distinction between what consumers or producers conform to

fsasaki: should we push for such a note?

<fsasaki> https://www.w3.org/International/multilingualweb/lt/track/issues/106

<scribe> ACTION: Stephan Walter to propose a note for the character encoding behaviour [I18N-ISSUE-246] [recorded in http://www.w3.org/2013/03/04-mlw-lt-minutes.html#action08]

ontology

<fsasaki> no further update, see discussion above

hanlding of defaults in test suite

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Mar/0045.html

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Mar/0044.html

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Mar/0046.html

kfritsche: the defaults in the storage encoding is a value, but the understanding of the default was that it was integrated later; we propose the behavior that would set it to utf-8 if no other value was present

Yves_: it's different behaviour than ordinary defaults

https://www.w3.org/International/multilingualweb/lt/track/actions/open

<fsasaki> action-390?

<trackbot> ACTION-390 -- Tadej Štajner to check tan examples and test suite with regards to presence of annotatorsRef -- due 2013-01-30 -- OPEN

<trackbot> http://www.w3.org/International/multilingualweb/lt/track/actions/390

<fsasaki> tadej: can be closed, done, changes are in github pull requeste

<fsasaki> close action-390

<trackbot> Closed ACTION-390 Check tan examples and test suite with regards to presence of annotatorsRef.

<fsasaki> tadej: did the check, all examples are valid

aob

fsasaki: anyone at CeBIT?
... we'll have some demos for ITS, thanks to Cocomore
... is the current schedule of calls ok for all?
... right now, we still don't have a right call time that works for everyone. let's try again after the review

- DRAFT -

MLW-LT WG

04 Mar 2013

Attendees

Contents