Ersatz Genius: now with 50% more crap!

About a week ago, Netizens were up in arms, rallying to the nearest forum/microblogging service/Intarweb journalist, complaining bitterly that Tumblr seemed to be down. And down it was.

Twitter was, of course, filled with the virtual gnashing of teeth of Tumblr users (while WikiLeaks-related topics were apparently being censored). While I was a little miffed, I wasn't sored that my blog was down; I was just sore that I couldn't get my latest entries.

My own Tumblelog from scratch in 24 hours

See, at that point, my entries up until 29 September, 2010 had already been backed up. As per my original post, I'd already begun fearing lock-in--in my post, I was referring specifically to Google, but I didn't want to be locked in with anyone.

So, back around October, I decided to sharpen my Ruby skills (which had grown rusty, due to my abhorrence of how Rails the framework had become the same as Ruby the language to many noobs). I pulled up Tumblr's API, opened an Emacs buffer, and started coding. (Afterwards, I rewrote the whole thing in Python, just for kicks).

Next, I set up an instance of lighttpd on a VPS I got cheap, then got PyBlosxom running on top of that, started writing a custom plugin that lets PyBlosxom render Tumblr posts plus a minor design refresh, and I was off!

You said 24 hours. It's been about a week since Tumblr went down.

Eh. I worked on my blog during my spare time. An hour or so every night, I'd be setting up a bunch of little plugins that I either wrote from scratch or existing ones that I customized a bit. Or tweaking some aspect of the design refresh. Or finding bugs, or missing features from the old blog that I'd want to keep.

All in all, not counting the time I've spent writing the latest two entries and setting up the server itself, it's been less than 24 hours, actually.

There are still missing features, though. Like search. I'm still looking into tying my blog into an instance of Xapian. It seems easy enough; just need the time to do it in.

Tumblr backup tool

Yeah, that. I still have it, of course, both the Ruby and Python versions. I suppose it'd be better off in the public domain (since it's a very simple script to pull down posts and group them by ID and by slugs). It's not up anywhere, though; if you're interested, drop me a line.

It's named keel, by the way, since I wrote it to minimized the hassle, should Tumblr ever keel over either temporarily, as it did a short while back, or--gasp!--permanently.

As for the PyBlosxom setup (i.e., the custom and customized plugins), I suppose, it'd be easy enough to put it up under a free software license at some point (if there's enough interest).

Undocumented features

There're undoubtedly bugs lurking somewhere. And there are still features I'm planning on putting in later. If you find something that bothers you, let me know so I can get right on it.

I'll get back to my regular mud-slinging in a bit.