Subtle little bug in a library costs me a few hours (Sigh...)

I've been working on merging the last month and a half of mainline changes into the restructured tree. It went very well for the most part, there were some changes from last month that caused a conflict or two, but nothing major, or so it seemed.

However, at some point during the merging I started seeing failures of the entire invoicing test-suite (in fact, I was seeing the failures yesterday, but I didn't recognise them then). Since I'd been in the middle of merging up changes in 2 different code-based I figured it had to be a problem in those changes or their interactions. Basically I was seeing failures to import, manifesting as either explicit import errors or failures to find codecs in the encodings space.

Spent a few hours trying to track that down. Didn't seem to be anything wrong with the code. It worked fine on the dev server, but not on the workstation. Finally realised that this test-suite might not have been run on the workstation in a few months (I've been doing most of my work on the laptop) and that it might be a problem with Python 2.5, which is only installed on the workstation.

A bit of Google magic and voila, Kid (the templating system) had a problem with it's import hook under Python 2.5 in November. Upgraded to the latest version and yay! The test suite passes. Still have a pending feature implementation and then a lot of testing to get done. I was hoping to be standing at this point around 9 this evening.


Comments are closed.


Pingbacks are closed.