[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: xml2pot
From:       Stephan Kulow <coolo () kde ! org>
Date:       2001-11-29 11:31:35
[Download RAW message or body]

On Thursday, 29. November 2001 11:47, Frederik Fouvry wrote:
> ,-- One day, you wrote:
>
> [...]
>
> | Maybe we should just change the entity from:
> | <!ENTITY kde-http "http://www.kde.org/">
> | to:
> | <!ENTITY kde-http "<ulink
> | url="http://www.kde.org/">http://www.kde.org/</ulink>">
> | (unwrapped, of course)
> |
> | I can't think of many places where we *don't* want the KDE website linked
> | to in the documentation, and it would remove the problem above too.
>
> The point was precisely to make possible among others their use
> in attributes (tags are certainly not welcome there ;-).  Another
> point was that the value in the above case is used twice, and so
> the entity should be used on both cases.  I entirely agree with
> Lauri's comment as far as only running text is concerned (I
> thought about it when adding the fully marked up entity, but
> decided against it for lack of generality).
>
> Now, if the problem is one caused by the parser (and possibly XML
> parser specifications), then I'm happy to go along with the
> proposed change.
To some extent we abuse ENTITies. As we can't evaluate the entity in the 
parser (otherwise the translator won't see any of them), we have to parse the
entity usages (I'm sure there is a SGML phrase for that :). But the SAX parser
reports tags as one - e.g. with an array of ids and values. And before or 
after the entities are reported. But there is no way a XML parsing program can
know when entities are used. 

The way we use ENTITies is only good for validating parsers (like meinproc
or db2html), but it's pure horror for non-validating ones. Keep that in mind.

Greetings, Stephan

[prev in list] [next in list] [prev in thread] [next in thread]