Author: Olivier Berger

I’ve been investigating the use of nXml (the XML editing mode for emacs), for the production of RSS news feeds.

All I can say is that it ain’t really obvious.

I wanted to be able to produce modern RSS formats, like RSS 1.0 (RDF Site Summary 1.0), to take advantage of the extensible nature of “real XML” RDF format, or if that wouldn’t be possible RSS 2.0 (here, RSS stands for “Really Simple Syndication”), which seems to be XML based but less strict, maybe.

Note that I’m not an expert in all XML and RSS matters, but I feel I have sufficiant basic knowledge to guess what needs to be done. OK, maybe not, actually… so I’d welcome any comments.

nXML uses relax-ng schema to validate (and also to provide other facilities like auto-completion) the XML edited in emacs.

I tried with a RSS1.0 file containing the RDF headers… but it looks like the only schema loaded is the RDF one, which won’t help nxml understand more markup that the RDF ore namespace. I though that maybe nXml would be able to load additional namespaces, which seems to be the goal of RDF in RSS 1.0, if I get it right … but actually it seems not.

I found out after some searching that someone named Joseph Reagle produced a relax-ng schema for RSS 1.0 (announced on a post to rss-dev@yahoo group) which seems to include elements of markup for the RDF parts of RSS1.0 files… but it’s pretty minimal… so what ?

Maybe it would be possible to find a RSS 1.0 W3C schema that trang could convert relax-ng … but I couldn’t find one 🙁 … too bad … or maybe it’s obvious… but I must say I don’t know enough of RDFS and other schemas…

Finally I may as well be using the RSS 2 schema (written by Dino Morelli, announced on a post on RSS2-Support@yahoo) in nXml, if such RDF cannot be handled by nXML ?

Thanks to Georges Silber who helped in this search, and mostly confirmed my findings.

Just updated amandalogp, which I hadn’t touched for long.

Amanda logs had probably been changed, and after the sysadmin upgraded amanda, my logs were no longer parseable.

Now I don’t get anymore errors, and the parsing is working again… until next upgrade ? 😉

The modifications are in the CVS.

I’ve improved a little bit my webpages generation with Docbook XML Website. Now, there are two versions of the pages : with the “tabular” look, and with a “lighter” look, for deficient readers, printing and so on…

FYI, I’ve arranged the backend (Makefiles and so on) to have a cleaner list of directories and files.

It’s not perfect though, and I think it may be worth investigating Silkpage, which seems mainly to enhance docbook website by providing an ant generation process… although I’m no big fan of Java… so, I’ll let one of my collegues evaluate it before migrating.

Interested readers may have a look at my Makefile and layout.xml files, and follow the links to the XML and XSLT and CSS files, which should all be available although not linked by default from the pages ;).

Next step will probably be to generate two set of pages, with a translation in english in addition to the existing french version.

Looks like it’s fairly easy to be able to use your own copy of OpenOffice.org 2 beta2 on Debian testing. OpenOffice comes only packeged in unstable at the present time, or for RPM-based distributions. But it’s easy to install the later version on a Debian testing machine, and it looks like it will run OK.

You will need to have alien installed to be able to convert the packages.

Here are the steps I applied :

• Download the tgz archive : English, x86, and your preferred mirror
• Extract it’s content in your home directory
• cd into it
• remove RPM/openofficeorg-testtool-*.rpm
• Convert the .rpm into .tgz. Apply the following script : for i in RPMS/*.rpm; do alien -t $i; done
• Extract the contents of the tarbals : for i in RPMS/*.tgz; do tar zxf $i; done ; it will create an opt/directory there.
• That’s it. you only have to run .../OOo2.0beta2_native_packed_en-US/opt/openoffice.org1.9.125/program/soffice to launch it.

I’m not sure everything will run smoothly but that should be enough to read the .odp documents sent by my collegues running Fedora Core 4.

Update 16/11/2005 : you may then remove the rpm and tgz files from the RPMS directory, which are rather big 😉

Following the previous installs on D505 or D510 machines, this time I tried to install Debian on a D810 machine of another collegue…

And that was a little bit painfull, and not everything works, but we have at least achieved a state where it seems to be usable in a desktop wired environment.

In addition to the steps I described for a D510, this time, other problems happened.

There is the problem of SATA support, which is solved in the 2.6.11 kernel from testing… but then you have problems since the network card is not properly supported (no tg3 driver). So unless you installed, when you were still in the 2.4 kernel, the necessary elements to recompile the braodcom driver (using bcm5700-source, I think), you’re in trouble 😉

But I’m afraid there would be problems too : the source package for the 2.6.11 kernels seem a little bit broken in testing…

So I took the definitive decision to install the 2.6.12 kernel from unstable, and finally everything (or almost everything) will work…

With testing and the 2.6.12 the testing package for the ipw2200 wifi driver won’t compile, etc… well other things from unstable may be necessary too… but I prefer to wait with testing until the 2.6.12 kernel comes with everything, to start over and reconfigure what’s still missing (wifi, bluetooth, etc.).

For a page describing the Debian installation on such a machine, you may refer to : http://tkrat.org/d810/