W3C home > Mailing lists > Public > public-i18n-its-ig@w3.org > October 2013

RE: Exclusivity and atomicity of local and global ITS

From: Yves Savourel <ysavourel@enlaso.com>
Date: Wed, 9 Oct 2013 23:09:33 -0700
To: "'Nathan Glenn'" <garfieldnate@gmail.com>
CC: <public-i18n-its-ig@w3.org>
Message-ID: <000a01cec57f$4a48da50$deda8ef0$@com>

There is no half-and-half: all the info of a data category instance (even the undeclared one) override the previous one.

"... Override semantics are always complete, that is all information provided via lower precedence is overriden by the higher
precedence. E.g. defaults are overridden by inherited values, these are overriden by nodes selected via global rules, which are in
turn overridden by local markup."

Or, in (better) your words: All of the provenance categories (org, person, tool, etc.) are considered as one when deciding what
overwrites what.

And it's for all data categories, not just provenance.

BTW Felix: there is a typo twice: it should be overridden (2 Ds).


From: Nathan Glenn [mailto:garfieldnate@gmail.com] 
Sent: Wednesday, October 9, 2013 10:48 PM
To: Yves Savourel
Cc: public-i18n-its-ig@w3.org
Subject: Re: Exclusivity and atomicity of local and global ITS

Thanks. For #1, I meant that the provRule element has a provenanceRecordsRefPointer attribute (see 8.11.2) and the tool markup
exists locally, so half local half global. Similarly with LQI, locQualityIssueType would be global and locQualityIssueComment local
(or the other way around, it doesn't matter). 
As for the local standoff winning over the global rule- what does that mean? If local standoff has person and org, and global has
tool, is the tool ignored? I understand that if local had person and org and global also had person and org that the local would win
out. I guess you could also ask, what is the granularity of winning out? Are all of the provenance categories (org, person, tool,
etc.) considered as one when deciding what overwrites what, or are they resolved individually?

On Wed, Oct 9, 2013 at 9:24 PM, Yves Savourel <ysavourel@enlaso.com> wrote:
Hi Nathan,
I think the scenario of your question 1 cannot exist.
You cannot have both a local reference to a stand-off annotation and a local LQI info. (or a global info with a global stand-off
As for a stand-off annotation and a rule: if the stand-off annotation is from a local rrference it wins over the global rule.
For #2 I think the rule applies, but the processor generates (possibly) some type of error if it tries to access the pointed
Just my 2 cents
From: Nathan Glenn [mailto:garfieldnate@gmail.com] 
Sent: Wednesday, October 9, 2013 7:55 PM
To: public-i18n-its-ig@w3.org
Subject: Exclusivity and atomicity of local and global ITS
Hello all,
I am wondering about of couple of possible situations in an ITS-decorated document that require information about the
exclusivity/atomicity of global and local markup:
1) If a rule of the same general category as existing local markup happens to match an element, but the rule and the local markup
give values for different exact categories, do they both apply? This question is only relevant for provenance and locQualityIssue.
For example, let's say there's a provRule that matches element X, and references a provenanceRecords element that contains person
and org information, and local markup on X specifies tool. Does the element then have ITS information on person, org and tool, or
does the local specification of tool erase any provRule matches? For locQualityIssue, a similar question would arise when a global
rule specified locQualityIssueType and the local markup specified a locQualityIssueComment. This one is specified with "in parallel
to local inline markup", so I'm guessing that both would apply.
2) If a rule has a pointer attribute that doesn't match, does the rule still match? Is it supposed to depend on what parts of the
rule are required attributes? For example say that the selector for this rule: <its:termRule selector="//term" term="yes"
  termInfoPointer="../def"/> matched a <term> element, but its termInfoPointer did not match anything. Does the rule still match?
Received on Thursday, 10 October 2013 06:10:00 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:11:30 UTC