Re: [gobo-eiffel-develop] xslt file splitting tool

[Eric B. <gob...@if...>]

Andreas Leitner wrote:
> 
> I would like to implement the xml file splitter we talked some time ago
> on this list. It should be really trivial - here is the problem:
> 
> currently my rng -> (framed html, docbook) xslt does require a version
> of xslt that has not yet been recommended and might undergo major
> changes. standard xslt 1.0 can only produce one output file from one
> input file. since i want to "backport" my rng xslts to version 1.0 I
> will need a tool that takes one xml file and splitts it up. it would
> expect a document like this:
> 
> <?xml verision="1.0"?>
> <ds:document_set xmlsns:ds="what namespace should we use?">
>         <ds:document file_name="foo1.xml">
>                 content of foo1.xml
>         </ds:document>
>         <ds:document file_name="foo2.xml">
>                 content of foo2.xml
>         </ds:document>
>         <ds:document file_name="foo3.xml">
>                 content of foo3.xml
>         </ds:document>
> </ds:document_set>
> 
> the splitter would be started like this:
> gexmlsplit input.xml
> and produce the three files "foo1.xml", "foo2.xml" and "foo3.xml".

Would that work to generate several HTML files from a single XML 
file as well? I guess that the HTML should be XML-wellformed for
that. I'm asking that because it is not clear to me whether the
doc for a given library, let's say $GOBO/doc/structure, should
be written in a single XML file from which several HTML files
are generated, or if we should stay with several XML files each
of which generating only one HTML file.
 
> Eric, do you want to have the backported rng doc xslts in gobo?

I guess so.

> If so I
> think we also need the splitter application to be an official gobo tool
> (like gepp, geyacc, ...).

Note that instead of a tool such as 'gexmlsplit', I would have
preferred that Gobo had a XSLT tool as I already suggested in
the past, and extend it with some forthcoming XSLT 2.0
functionalities such as the generation of multiple files.
But I guess that writing a XSLT tool is not an easy task.
(I know, there are already plenty of very good XSLT tools
such as Xalan, but I still have this ideal view where the
Gobo package would be self-contained and only depend on the
Eiffel compilers, and possibly their backend C compilers.)

> Of course if none is interested in the rng docs, i won't even try (;

I'm interested in RNG docs, but I don't see it as THE doc. In
the same way Eiffel flat-short forms are not THE doc of a library
(such as $GOBO/doc/structure) but part of the doc, I think that
information extracted from RNG descriptions should be part of the
doc of 'gexace' and 'geant', but the doc should also contain
more global information. For example in the doc for 'geant'
Sven could explain how inheritance works, as he did in a message
to this list last month. I don't think that this kind of 
information belongs to the RNG file.

===

Since we are speaking about docs, I'd like to inform you that
I'm almost done with the planned modifications related to the
unicode string classes. When that is done I will still have
to check that my modifications in the regexp library work with
all compilers (I want also to come up with a common interface
between the PCRE regexp classes and those already in
$GOBO/library/lexical/regexp). When that is done and when
Franck is done with his modifiations in the XML library,
I think that it will be time to release this long-time promised
pre-release version of Gobo. That way third party projects which
have been using the CVS version of Gobo until now will be able
to tell their users to download this stable pre-release package
instead of chasing the moving target with the CVS version. Then
we should put all our efforts to work on the doc, guidelines,
test cases, etc. to make the official release of Gobo look as
professional as possible. So Andreas work on RNG docs belongs
to this effort.

-- 
Eric Bezault
mailto:er...@go...
http://www.gobosoft.com

 
______________________________________________________________________________
ifrance.com, l'email gratuit le plus complet de l'Internet !
vos emails depuis un navigateur, en POP3, sur Minitel, sur le WAP...
http://www.ifrance.com/_reloc/email.emailif