Re: ITS rules for OpenDocument

Hi Rodolfo,

Rodolfo M. Raya さんは書きました:
> On Tue, 29 Jul 2008 19:13:00 +0900
> Felix Sasaki <fsasaki@w3.org> wrote:
>
> Hi Felix,
>
>   
>>     * an implementation of "within text" which was necessary to get the
>>       OpenDocument segmentation right
>>     
>
> Segmentation is set at paragraph level, not sentence level. Is that
> correct?
>   

Yes. The sentence level is not marked up. I think your tool is looking 
into the content to determine the sentence level, e.g. "NOTE:" in your 
XLIFF output is a separeate <source> unit, although only part of a text 
sequence in a <w:t> element. I am working only on the markup.

>   
>>     * a now worky docx file, please check
>>       http://www.w3.org/International/its/its-translate-decorator/example/alice-in-wonderland.docx
>>     
>
> It is fine. Word 2007 can open it.
>
>   
>>     * the updated content of that file, see
>>       http://www.w3.org/International/its/its-translate-decorator/example/alice-in-wonderland.xml
>>       . Rodolfo, could you use that file to re-generate your XLIFF file
>>       to make comparison easier?
>>     
>
> I used the .docx file to generate a new XLIFF. You can download it from
>
>    ftp://charmed.maxprograms.com/pub/alice-in-wonderland.docx.xlf
>
>   
>>     * the re-generated XLIFF files
>>       http://www.w3.org/International/its/its-translate-decorator/example/xliff-file-alice.xml
>>       http://www.w3.org/International/its/its-translate-decorator/example/xliff-file.xml
>>     
>
> Some comments:
>
> 1) the official extension for XLIFF files is ".xlf". Don't use ".xml"
>   

OK.

> 2) use a real language code in the generated XLIFF files. Set it to
> "en" (English) or anything else, but not to "tbd". XLIFF editors are
> able to validate language codes and complain if you use an invalid one.
>   

OK.

> 3) In "xliff-file.xml" you declare source-language="en" in the <file>
> element and then use "tbd" in all <source> elements. This is
> inconsistent.
>   

OK.

> 4) There are too many inline tags in "xliff-file-alice.xml".  Tags that
> contain the whole segment can, in most cases, be excluded. For example,
> if the whole segment is enclosed in <bold> tags, you can put the tags
> in a skeleton and store clean text in the <source> element. Tags that
> appear before or after the segment and don't affect the text can be
> exluded from the segment too. 
>   

As I'm not an XLIFF expert, this is hard for me to judge. Maybe we 
should do what Christian proposed at
http://lists.w3.org/Archives/Public/public-i18n-its-ig/2008Jul/0033.html
that is, to separate the decorator from the XLIFF generation , and leave 
the output to real experts like you. What do you think?

> 5) <ph> element has a required attribute: "id". It is missing in your
> files.
>   

OK.

Many thanks for the feedback,

Felix

> Best regards,
> Rodolfo
>

Received on Wednesday, 30 July 2008 11:15:39 UTC