Regarding a), I’ve now been enlightened after discovering the joys of virtualenv (+ virtualenvwrapper of course), easy_install and now pip. Now pip – being the newer of the two and being virtualenv aware – is obviously better, but you still have to use easy_install to get it in the first place. virtualenv makes managing multiple projects with multiple versions of libraries (i.e. Django) sane, but overall I still get twitchy when installing applications – I miss apt-get and Synaptic!

Yes, I know about Fink etc., but so far the experience with those has been suboptimal.

For PostgreSQL, I went with the installer for 8.4 maintained over at EnterpriseDB. However when I tried to install psycopg2, I got an unhelpful message from both easy_install and pip about a file not being found, but not which file. A quick Google around led my to this post which pointed me in the right direction. The trick is to provide pip (or easy_install) with the bin path for PostgreSQL. Therefore the magic incantation for me was:

$ PATH=$PATH:/Library/PostgreSQL/8.4/bin/ sudo pip install psycopg2

As an aside, I’ve been using sudo for years (and years), so why do I keep forgetting to use it with OS X?

All working now!

Today I finally got around to finishing the first batch of changes: namely Django 1.0.2 and moving from using a text file based word list to a database (which was already used for storing the URLs). This improves performance and opens up the possibility for a whole other bunch of enhancements (including using your own words).

At the same time I added some improved error handling & error messages, and the ability to ’shortcut’ previewing a link by suffixing a hyphen (-) to the word e.g. http://linkpot.net/recursive-.

If you use Linkpot, please go kick the tyres and let me know if you find any problems. If you don’t, why don’t you?

I agreed with Stuart that it should be released regardless of demand, but it really did need some cleaning up first!

So, without any further ado, pomp, or ceremony I give you m2wp.py!

…

Oh, you actually want to see the code?

• If you just want the file, you can grab it here
• If you just want to see the the code, you can see it here
• If you want to be all modern and distributed, you can grab the code using Bazaar like this*:

  bzr branch http://labs.schwuk.com/m2wp

* Why aren’t I using Launchpad? Because I seriously doubt there will be any further development of this. However if people want to submit patches, they are more than welcome to and I will publish them in my repository.

A couple of caveats:

• It assumes you’re using a PostgreSQL database. If you want something different, add it.
• Due to the wonders of WordPress, you can’t have tags and categories with the same name, but you can in Mephisto. You’ll need to fix this once you’ve imported into WordPress.
• The script is released under the GPL, and all that that entails.

After what seems like an eternity (but is actually just over a year) I’ve switched the backend of this site from Mephisto to WordPress. The main reason for the switch was my overall dissatisfaction with Mephisto coupled with its lack of development (or a least stable releases) along with the fact that I was maintaining a Ruby on Rails production environment for one application. Mephisto frustrated me in many ways, from its inability to handle archives/pagination to not being able to accept pingbacks to not working with external clients. All of these were fixable, but I really didn’t have the patience or the time.

Here is my tale of migrating from Mephisto to WordPress, and how I achieved it.

I was also going to write my own ‘blog’ software in either Django or Rails (apparently this is the mark of true developer), but again I’ve got better things to do and with a couple of projects I’m involved with utilising WordPress it was less “eating my own dogfood” more “eating the dogfood I would be serving to others”.

As an aside, the dogfood analogy really is gross, but it works.

I initially despaired with migrating from Mephisto. Although there are a couple of scripts to migrate to it, as a fairly niche product there are zero scripts I could find to migrate from it.

So I started writing my own.

My original plan was to simply massage the data between the two databases, but with my Mephisto using PostgreSQL and WordPress being restricted to MySQL that was never going to be easy. My second plan was to dump the data out into some interim format, then inject it directly into the WordPress tables. At this point I started playing around with WordPress itself, and discovered its ability to import data from a variety of formats, including its own WordPress eXtended RSS (WXR) format – a format that isn’t too popular. My third plan was formed – I would craft a Python script that would generate a WXR file from my Mephisto database, then let WordPress handle the import for me.

Getting the data into the right format was fun, but xml.dom.minidom served me well. Importing the data proved a bit more problematic due to the fact that the import script in WordPress doesn’t do XML at all, but rather Regular Expressions. A couple of tweaks to my script and the file was ‘well formed’ as far was WordPress was concerned.

The next import problem was caused by tags and categories. These are separate in Mephisto (and so names can be duplicated), but not so in WordPress. A bit of manual cleanup was required post-import, but the Tag Managing Thing plugin made this less of a chore and let me optimise my tags and categories at the same time.

The script is by no means production ready, and is currently tailored for my preferences, but if there is enough demand I will Open Source it.

Of course no upgrade is without its problems. All permalinks and tags should continue to work, but categories have been moved into the /category/ path. Feeds should be mostly working (I’ve manually updated the two planets that use category specific feeds, and I’ll keep an eye on my stats to see if any others are being used). If you find something that is broken let me know. The other obvious difference is that my design (for those of you that don’t live in Google Reader) has gone. I’m working on a new design, but until then K2 (with some tweaks) will suffice. I’m sure you’ll cope without the ads…

Of course WordPress gives me some new toys to play with. Trackbacks now work, and I’ve enabled OpenID for those that use it (i.e. me). I’m still using Akismet for comments, but I’ve also added a captcha – I’ll see how that goes.

Now Bruno Bord (who’s basically French) has come to the rescue with the PyRoom project. I may have to get involved.

Today marks the last day of my employment with CSC. After 9 years (originally a freelancer, then the last 4½ years as a permanent employee) it is time to move to pastures new.

On Monday I will be joining Canonical as a developer on Launchpad (any comments about (open|free)ing of said product should be directed to Matt).

The best bit? Getting to work with a passionate group of people who believe in what they’re doing. The worst bit? Having to explain to non-technical family and friends what Canonical do. The best bit of the worst bit? Starting to explain it to the schwuklets, realising who I’m talking to, tell them they help make Ubuntu and schwuklet #2 (who is only 5 years old) knowing exactly what I’m talking about!

Of course every silver lining has a cloud, but you have to take the (dog) rough with the smooth. (I’d put a in here, but you know how much he likes those…)

In a very short space of time a working result was achieved, but said associate no longer had a requirement for due to a change of platform. Rather than let it languish, I started adapting it to be more generic and removed some additional (and not fully implemented) functionality to create the WebDev Control Panel.

A project which promptly vanished into the depths of my hard disk and was mostly forgotten about. I fetched it out from time-to-time and tweaked a bit here, changed a bit there, but overall it sat collecting dust. I fully intended it to be released as Free software, but was loath to release in its current form.

Jumping forward to a couple of weeks ago, I needed to re-acquaint myself with Python, so I dusted off the project and started polishing it. I made the code more object-oriented; I converted it from a fixed list of services to a dynamic one (controlled by a config file); and finally I made it ready for translation.

Finally the time had come, and I could find no decent reason for putting it off any longer. I had to release it.

So, after that long introduction, here is the WebDev Control Panel. Enjoy.

It is an introduction to Rapid Application Development with PyGTK, and I’m very pleased with the result. Of course real feedback (instead of my own ego massaging) is always welcome.

I am intending to follow this article up with a series on Python development (with a twist) – further bulletins as events warrant.

In the mean time go and pick up a copy!