Trial Releases

I thought I’d have to wait another week or two to feel good about the next major (slightly backwards incompatible) release of Dist::Zilla. It looks like I should be able to release it this week.

I’m feeling pretty good about progress so far, especially after today taking an axe to the InstallDirs plugin, which was a bug ugly mess. Here’s the big board:

I don’t think I’m quite done with my basic tests, but I’ve gotten to the point where I need to fix some other features first. The yak stack is actually sort of interesting, so I’ll explain it:

I’ve added a bunch more tests to Dist::Zilla as I try to work through my checklist of plugins to cover. Some have been easier than I expected, but I’ve found myself in a bit of a predicament.

“Hey, look, I wrote some more unit tests!” is about as fun to report as it is to do, but so far, so good. The Dist::Zilla::Tester system I put together has been working pretty well, and I’ve made a few tweaks as I go. I’d originally meant to write one test per plugin, but for some of them it’s just much, much simpler to combine them to have two or three plugins per test file. If this leads to problems, I can split it up again later. For now, it’s looking good.

I’ve made a rough checklist of all the things I need to write proper tests for. This will only get me unit tests, and integration tests will then be an ongoing effort, due to the many possible configurations I will want to test for.

Guess what I worked on today in Dist::Zilla? Logging! I’m beginning to think I may have a logging-tweaking addiction.

In short, there was just too much copy and paste. It wasn’t really trivial, either. Here’s an example of code that appeared in both Pod::Weaver::Role::Plugin and Dist::Zilla::Role::Plugin:

The last 24 hours or so have been really busy. Today was Martha’s third birthday (wow, can you believe it?) and I’ve had a number of incredibly annoying bugs at work. I won’t get into the details right now but one of them was an MSIE bug that I solved by \uXXXX-only double-JSON-encoding a structure. The other was fixed by disabling Etag headers. Blaaaagh!