Re: LSB SGML/XML appendix: finding the DTD
/ Eric Bischoff <e.bischoff@noos.fr> was heard to say:
| > You don't need to modify every tool get solution 4 to work. Most Java
| > based parsers use SAX and/or JAXP. Some common tools let you specify
| > resolvers on the command line.
|
| It's probably a point I didn't get - can we provide a resolver that would
| work for every tool? Or do we have to provide a patch for Saxon, a patch for
| libxml, a patch for Xalan/C++, etc?
It's going to vary a little bit according to the environment.
For Java, we can provide a class that will work for everyone. It may
require a little tweaking around the edges (command line parameters
and the like) to get it to work for each tool.
For C, I hope that we can get Arnaud to include it in libxml. That
should cover most C tools. I'm not sure what we'll need to do to get
it working in expat.
| > For other tools, you'd "only" have to
| > use a SAX/JAXP parser that did resolution.
|
| and what about DOM?
You usually parse to build the DOM, so I'm not sure it's relevant. But
the resolver will work for any access that goes through the URIResolver.
| The question is: do we nevertheless provide a framework for people to do it
| cleanly, if they really want to go this way? All document and tools packages
| would be using the same method and thus be interoperable. Or do we say that
| it's encouraging bad habits?
If people want to do it, that's their business. I don't think
distributions should encourage it.
| One big problem if you allow both methods is that every SGML/XML-aware
| package has to provide both the symlink for B3 and the post-install script
| for B4 :-((.
C'est la vie. TANSTAAFL.
Be seeing you,
norm
--
Norman Walsh <ndw@nwalsh.com> | Debugging is 99% complete most of the
http://nwalsh.com/ | time--Fred Brooks, jr.
Reply to: