So while I’ve got the tinderbox turned off I’ve been taking care of a few different QA duties that I’m probably not supposed to do but I’m sure to help Gentoo. I’m actually pretty sure that this kind of tasks might actually be even more interesting for users than what I’ve been doing with the tinderbox.
While the tinderbox’s main goal is to be able to find the broken software that is in the tree, this usually produces just a lot of work for other developers (bugs to fix) and a few extra side-effects like identifying smaller QA violations and some very broken package that I have been last riting and that will be removed over the course of the next two months.
On the other hand, the manual analysis I’ve been doing tonight aims to check the actual stuff that is added to the tree, like binary files or big files directories. For those wondering why I’m on a crusade against binary files in the tree, I have to say that first of all, CVS makes it difficult to handle binary files, and this makes them unsuitable for being added to the tree. Additionally, binary files in the tree often mean there is something else broken with the packages: compressed big patches (that still keep big) and huge, messed up files directory with unused content, and stuff like that.
I’ve been able to shave a few kilobytes off the tree by moving a few files on the mirrors and removing the big files from the tree; but I’ve also started sending last rites for the packages that have this kind of issues and I don’t see as being ready to be fixed sometime soon. Interestingly enough, it turns out like there is enough cross-over between the packages that fail, that have QA issues and that are polluting the tree with too-big files directories and so on.
So please don’t get mad at me again if I masked for removal a package you use: if you want to keep it in the tree, please get it fixed.