When I decided to take a break to write, I really should have been stricter with myself and only started one book. Instead, I started three. While multitasking and context-switching may be a good thing in the fast-paced, high tech world of the Internets (and I'm not convinced it is), there's a chance it's really unhelpful when you're trying to write a book.

That said, I have at least finished one of the books:

How to Catch a Cold

It's a pencil and watercolour illustrated story about a boy's attempts to avoid school by catching a cold.

Now to move onto the second stage of book writing: wallpapering my bedroom with rejection letters.

Really, it looks worse than it is.

For reasons of top-secret international security, I can't go into details but I'm back in hospital. pic.twitter.com/wWgaU100vJ

— Simon Madine (@thingsinjars) April 27, 2014

I'm actually just getting my tonsils out but seeing as it's only really teenagers that get their tonsils out, I'm trying my best to pretend there's a parasitic fungus attacking the planet rendering everyone helpless and my blood is the only cure.

Yes I'm bored.

Things will be quiet here for the next few months. I'm taking a break from tech writing for a while to see if I can write something else.

You might remember the last time I did this when I wrote Explanating and tried the (admittedly, not very successful) read-then-buy experiment.

I'll update again when I've got something to show for myself.

Thanks to some great work by Daniel Wabyick and his team, Hardy has had a bunch of improvements over the last few weeks.

The biggest change in this version is that, if you have GraphicsMagick installed on your machine, Hardy will use it for native image diffs but fall back to the built-in method if you don't. The current image diff technique involves creating an HTML page with a canvas, opening that with PhantomJS, loading the image into the canvas and using imagediff.js to calculate the diffs. It works everywhere PhantomJS works but it's slow. Daniel benchmarked the difference and it's a huge performance gain if you rely on image diff tests.

There's also some minor improvement around logging and the cucumber report format but I'll write about them later once I've had a chance to update the Hardy website.