Archaeogeek is now migrated to Octopress! There are a couple of things still to sort out, such as some old broken links, a non-default theme and comments on older blog posts. Hopefully the pharma hack that has been polluting my google search results for a while now will disappear as the site is reindexed…

There are a few stuctural changes. I’m no longer asking people to register for downloading Portable GIS for a start. I had planned to use a mailing list to notify people of updates but never got much further than creating a table of names. It was useful for counting how many people download the file, but I can manage without knowing precise numbers. I’m no longer offering a contact form- there are multiple ways in which people can get in touch if they need to, and the most common thing people used it for was spamming me with SEO requests anyhow!

It has been a fun process- and thanks to James Fee for giving me the impetus to go ahead!

The rest of this post comprises some notes on actually doing the upgrade, not-so-subtly giving me a chance to try out some markdown syntax!

Comments have been the biggest pain actually. The default (only?) option with Octopress is to use Disqus, but I couldn’t get the wordpress export to work, and the manual import on the Disqus site seems to be taking a very long time! Also my old posts seem to have comments switched off in the markdown, so at some point I’ve got to figure out a pain-free method of switching all of those back on without having to do it manually, and without re-posting all of them! This will probably involve a python script of some kind.

I did read a number of blog posts about exporting from wordpress, which suggested the process was quite hard, but luckily it went OK. I used the built-in wordpress export tool to download the blog as xml, then the Exitwp tool from here. This recommended using xmllint to check for issues, but it was quite hard to understand the results. In the end I just ran the script anyhow, which reported a single error (phew). I went back to wordpress, found the offending post, used the automatic “close open tags” option, re-exported the xml, re-ran the exitwp script, and all was fine.

Another potential gotcha was images, which in wordpress are in a folder called “uploads” and then separated out by year/month folders. Luckily I didn’t have very many embedded images, so those could be fixed manually but if you had a lot then you’d need some sort of automatic process to fix these, or replicate the same structure in your new site.

Finally, I decided to use ssh and rsync to deploy to my remote site, which again was remarkably simple. I had to ask my Web Host (A2, who have been superb recently) to enable ssh and rsync on my site, used ssh-copy-id from my linux pc (in windows you’d need to do this manually using putty or similar) to copy across my public key, and then set up rsync in the octopress config just like the documentation says.

One thing I was really worried about with Octopress was the time it would take to re-generate the site each time I made a change. This currently takes about 2 minutes, which I can cope with, unless I get considerably more prolific!

Still to do: