W3C home > Mailing lists > Public > public-i18n-its-ig@w3.org > June 2008

Re: Local support for "Elements Within Text"

From: Felix Sasaki <fsasaki@w3.org>
Date: Sat, 14 Jun 2008 01:34:37 +0900
Message-ID: <4852A19D.3090201@w3.org>
To: Yves Savourel <yves@opentag.com>
CC: public-i18n-its-ig@w3.org

Hi Yves, Ingo, all,

Yves Savourel さんは書きました:
> Hi Ingo,
>  
> Thanks for following up on your input of last teleconference,
>
>
>   
>> It may be beneficial to support the ITS data category 
>> "Elements Within Text" at the local level.
>> Supporting this at the local level would mean that a simple 
>> parser (which does not have to deal with XPath statements) 
>> could be written.
>> Needing a complex parsing procedure may put some people off 
>> using ITS.
>> Segmentation information is important to many editing tools, 
>> and having this at the local level is of benefit to these 
>> tools (if and when they support ITS).
>>     
>
> Obviously, one of the reasons why the current version of ITS does not have support for local Elements Within Text rule is because we
> found only very rare cases where an element would have to change its 'within text' property inside a document, as this would means
> the element has suddenly different semantics. And the examples you gave indicate no differences on that aspect.
>
> The main issue I see with your proposal is that the rational for adding a local Within Text rule seem solely based on making things
> easier for the ITS processor.
>
> So I've tried to imagine how this would help:
>
> Having a processor that support Winthin Text only at the local level would means it can only process documents where every single
> element that is either 'within text' or 'nested' would have to be marked up as such. Every bold, italic, every span-like elements...
> This seemed a bit un-realistic to me.
>
> Then I thought about default attributes in DTDs.
> I guess one could argue that a DTD could define the default its:withinText attributes for each element where it is required and this
> would result on a document where each element has that information at the attribute level, and this not being related to any real
> local change. It would be simply a different way to pass the 'global rules' to the document. And this would allow an ITS processor
> supporting only local rules to handle ElementWithin Text.
>
> I am not sure this is a scenario I would recommand to anyone: Default attributes have their own set of issues. But it is a possible
> case.
>
> Now, I'd like to hear other people's opinion on this. This is an interesting case.
>   


Sounds like dangerous ground to me, since you don't have that 
information available without a DTD. I think what you really want is to 
have less expensive global ITS processing, that is not requiring the 
full power of XPath, but having e.g. just a list of elements and / or 
attributes. Maybe this is the general topic of this thread and the other 
one on reduction of XPath to some axis.

Felix

> By the way, It reminded me that an implementation of an ITS processor had to handle such case for other data categories like
> Translate: That default attributes should be overriden by global rules, while non-default attribute should override the global
> rules. Thanks for the reminder.
>
> Cheers,
> -yves
>
>
>   
Received on Friday, 13 June 2008 22:10:07 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:11:27 UTC