Re: DOCBOOK-APPS: Catalog dor DocBook and URI for the XSL stylesheets

[Mark Johnson <mrj@debian.org>]

Hi Guys, 

I'm almost through my 2-month backlog of email. Ugh.

Anyway, this thread that Daniel started hits right on an issue that
I've been thinking about -- persistent identifiers for the docbook-xsl
stylesheets. 

It sounds to me like the plan is to have, say, the xsl stylesheets at
a persistent URL:

   http://docbook.sourceforge.net/release/xsl/{$VERSION}/...

And since Norm also keeps the current version at

  http://docbook.sourceforge.net/release/xsl/current/...

the latter would work fine for debian as the URI to map to the local
installation via the 'rewriteSystem' element (debian doesn't put
version numbers [yet, anyway] in the stylesheet directory trees.)

Am I on the right track here in interpreting the thread?? 

I'm not sure I understand Daniel's point about catalog/directory
structure templates, though. If he means that 
  (1) we all agree on the same relative sub-directory structure, and 
  (2) that the agreed upon structure mirrors that of the 
      upstream distribution,

then I wholeheartedly agree. Point (2) is actually part of our
proposed policy for debian. On the other hand, based on the lsb-sgml
spec discussions, it seems very unlikely that the various
distributions can agree on a common sub-directory structure. 

Maybe this one will have to be decided via the 'effective practice'
model, where the least problematic solution is the one most widely
adopted. I dunno. 

BTW, this discussion is very timely for me, as I'm in the process of
packaging Norm/Sun's resolver classes. I also have to think hard about
an XML Catalog system/structure to propose to the debian folks. The
catalog structure in the current (stalled) draft of the SGML LSB spec
has some problems that I'd rather not propagate. (Sounds like Daniel
is going through the same process in implementing his libxml
stuff. Daniel: BTW, do you ever come to RedHat Central in North
Carolina?  I'm 10 minutes away from the place... It'd sure be nice to
get some face time on this stuff.)

Finally, I, too, have experienced funniness in http connections to
sourceforge AND to oasis. But that won't really be an issue if users
have local installations of this stuff and use catalogs. Isn't that
the whole point of having catalogs, anyway?  The cache mechanism
(though very, very cool) simply won't be the right solution for some
users, esp those with sporadic connectivity.

One Last Point:

I may be able to offer to have Duke host something like a
docbook-everything-release mirror, as they're big on establishing an
edu presence on the XML scene. Since they've enthusiastically agreed
to become the new home of the local TriXML site & meetings, and I'm
about to propose that they make a stronger committment to support the
LSB-sgml-xml spec work, I could also request that we host a mirror of
the various docbook-related releases.  (Heck, I own the
docbookstuff.org domain anyway...)

Does such a setup sound like a good idea to y'all? 

We could get a domain like 'docbook-home.org', etc. 


I do believe this post satisfies my email fix, I feel much better
now:-)

Feedback, please.

Cheers,
Mark

On Friday, August 24, Kevin Conder wrote:
> On Fri, 24 Aug 2001, Norman Walsh wrote:
> 
> > They do now, actually,
> > 
> >   http://docbook.sourceforge.net/release/xsl/{$VERSION}/...
> 
> 	Please consider using another site. Sourceforge uses a strange 
> port-redirection scheme that doesn't work well with corporate firewalls.
> For example, I'm only able to see Web pages on ports 80 and 8080 at work.
> I doubt I'm the only person who has problems with SourceForge...
> 
> 
> === Kevin Conder, kevin@kevindumpscore.com
> 
> 
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.oasis-open.org/ob/adm.pl>

-- 
_____________________________________
Mark Johnson
Duke Physics        <mark@duke.edu>
Debian SGML         <mrj@debian.org>
Home Page:          <http://dulug.duke.edu/~mark/>
GPG fp: 50DF A22D 5119 3485 E9E4  89B2 BCBC B2C8 2BE2 FE81