[XML-SIG] XML DTD for RPM spec and PO files

[Nicolas Chauvat]

> What do you gain by doing that? For the existing formats, all kinds of
> tools exist. For the XML equivalents, nothing exists.

My idea is to use XML as a common format and reuse the xml tools to do the
same processing for different document that I received in different source
format.

In french I'd call it a "format pivot". I suppose I could translate it to
"hub format".

I know that tools exist for other formats. But I'd like not to have to
install, use and know of as many tools as (format, processing) couples I
have to deal with.

> > Then I can store all these files and there translation and use the same
> > tools (xmldiff and XSL transforms) to deal with new versions, detect pa=
rts
> > that changed, produce reports and even generate the original format (XM=
L
> > -> po or XML -> spec).
>=20
> Why do you want to use, say, XSL, on a po file? The typical output
> processing of such file is into a binary .mo file, which is structured
> for efficient access at run time. I very much doubt you can do msgfmt
> in XSLT.

That is not what I meant. My idea is to archive source documents in XML
format, do version diffing and checking with XML, maybe write the source
directly to XML, then turn these documents to the proper format when
needed.

For .po files, it would be something like :

XML-formatted .po -- XSL --> .po -- usual po generation tools --> whatever

For .spec files, it would be something like :

XML-formatted .spec -- XSL --> .spec -- RPM --> package

For man pages, it would be :

Docbook -- XSL --> man=20

or Docbook -- XSL --> man source -- groff (?) --> man

Disclaimer: I don't know much about source po and man usage and generation
I probably used the wrong tool names above.

> Likewise, to combine revisions of catalogs, I doubt any tool
> would be as good as msgmerge, with support for fuzzy messages and all
> that.

Agreed. That's definitely an issue with .po files. I suppose one could
write an XSL extension that does the same, but is it worth it ?

--=20
Nicolas Chauvat

http://www.logilab.com - "Mais o=F9 est donc Ornicar ?" - LOGILAB, Paris (F=
rance)