The assorted finds of Artefact Publishing

Switching to WordPress

So, I was going to switch over to using WordPress to run this blog. After all, it’s open source and has an active community. Well, I haven’t done it, and here are some reasons why:

Now, don’t get me wrong, most of these are also problems with Movable Type. But in order for me to go to the effort of switching (and I put in a chunk of time this afternoon on it), I want something closer to perfection. So I shall wait awhile. All was not lost, however, as part of trying out the migration meant I got to clean up lots of the cruft in my entries, which I imported back into Movable Type.

Posted by jamie on May 15, 2005 18:57+12:00


Clearly you are just going to have to write your own blogging system.

Posted by: Michael on May 17, 2005 12:06+12:00

I am.

Python + CherryPy + Sqlite + Markdown + xyaptu + all the goodness of pseudo-continuations for state.

Mmmmm, continuations.

Unless my Haskell book arrives soon. Then I might just have to start all over again… should be ready in time for 2007!

Posted by: stephen judd on May 17, 2005 14:12+12:00

I’ve heard good things about CherryPy, and I’ve just recently been using Markdown for the manual for Amara. However, any blog system I work on will not have anything to do with minimizing markup. None of this XHTML for me, either, thank you very much — I’ll write my entries in DocBook or TEI. Stick Apache Cocoon in front to transform the source into whatever, and away we go.

Publishing should be as simple as putting a file in an appropriate directory, with a cron job that harvests topic map information from the file. Comments are created as files too, with the submission process kicking off the harvester, and trackbacks are handled similarly (or just going straight into the topic map without having any other representation).

Yeah, I’m going to be writing this really soon.

Posted by: Jamie on May 17, 2005 21:45+12:00

Rendering out to static files is a very sound idea, which I shall now steal. In my schema, posts and comments (and potentially other content types) are morally equivalent, live in the same table, and are composited into a page. However, I want a web-based front end (for remote access) and I prefer relational data stores. Of course, I'm biassed out of fear. DocBook is scary. Cocoon is scary. Can't sleep. XSL will eat me.

Posted by: on May 19, 2005 11:29+12:00

Cocoon would be a lot less scary if it was well documented (a common lament). I’m sure you’re not serious about DocBook being scary, given that compared to TEI it’s a simple and limited tagset.

Me, I’m just glad DSSSL isn’t waiting under my bed (it is dark; you are likely to be eaten by a sosofos)....

Posted by: Jamie on May 19, 2005 18:11+12:00

DSSSL looks like the friendly waggy puppy Scheme to me, so obviously our tastes differ.

I've been so intrigued by your committment to topic maps that I have started reading up on them. They're neat! I can see all sorts of lovely site-nav generation based on a topic map. I can't help but be reminded of an old-skool network database, however :-)

Posted by: stephen on May 20, 2005 19:04+12:00

Looks like Scheme because it basically is. It’s not that that scares me — it’s the sheer volume of stuff that needs to be specified in order to output anything useful, and the fiddliness of same. Perhaps it’s not so bad in practice, but it sure looks like you’d want to either use someone else’s work and tweak, or create your own house style and never deviate from it.

Glad I’ve given you something new to play with — it’s certainly a pretty cool thing if you put the work in. I’ve just started reading Elaine Svenonius’s The intellectual foundation of information organization which should provide plenty of insight I can apply at the NZETC with regards to our ontology.

The idea of topic maps is of course a relatively simple one; one of the areas where it really shines is in merging topic maps, which is what you’re going to do a lot of if you have plans to take over the world. And really, who doesn’t?

Posted by: Jamie on May 20, 2005 19:23+12:00

XSL is also scheme, just with a really bad surface syntax. And actually suprisingly powerful, despite the paucity of useful string manipulation functions. Me, I'd think about using PLT Scheme -- real honest to goodness continuations...

Posted by: Michael on May 28, 2005 11:30+12:00

Hey, Michael!

XSLT 2.0 (or rather XPath 2.0) has better string manipulation, which will be a blessing, but indeed XSLT is a useful and powerful tool even as is. The problem areas that I run into at work are not to do so much with XSLT as with differing content models (converting TEI to HTML, with paragraphs containing other paragraphs and lists etc into separate parts) and the constraint of non-overlapping hierarchies (leading to the use of milestone elements and other hackeries).

Posted by: Jamie on May 28, 2005 12:06+12:00

i just want a pretty site, that's free

Posted by: sue on July 2, 2005 18:25+12:00