Testing troubles

drhyde on 2007-03-27T10:18:47

CPAN testing takes quite a lot of time - mostly hitting enter when modules using non-standard installers annoyingly ask "do you want to install my pre-requisites", but also checking by hand before sending a FAIL result that it's sensible to send it, and that I've not already sent that particular result before, figgering out what the heck to do when a module asks about obscure libraries and applications, or how to configure a module. Then there's optional extra modules to install, where I have to be careful to not choose modules which I know fail their own tests, and so on.

It also takes a lot of time because I often fall into the temptation of finding the bug and sending a patch. Some such patches have been accepted by authors, which pays off the karmic debt I owe from wishing painful, messy death on half the denizens of this fine city. Others have been *requested* by authors, sent, and seemingly ignored. Grrr. And then there's the bugs that are hard to track down. Last night I decided to spend just a few minutes looking at why a module wasn't working on perl 5.6, and the next thing I know, an hour and a half had passed. Bah.

I can at least partially solve the first problem. As of now, I'm dropping 5.8.8 on Linux from my list of testing platforms. I'll keep 5.6.2 on Linux and 5.8.8 on OpenBSD so between them those two should catch the same bugs. That'll free up the time for me to add some other platform - once I can get perl to build, of course.