Yes I know I start to get boring on this matter, but since lots of users lament about the "size of the tree", I think this is worth stressing a bit.
I'd say that the file.size check should be splitted in two parts: >10K is a warning, >20K a fatal error. This way we can actually have some sane limits on the system, I'd say.
Zac please give priority to this as soon as you have some time; I just found a 41K patch _added to the tree_ (for Python 3 compatibility, no less). We're reaching ludicrous situations.
Three years ago there was a discussion about patches.g.o where patches would also be archived. Maybe we should restart the discussion.
If the ebuild is removed the patch would be lost after two weeks.
True but we got a cvs module for patches (or git repositories for those disliking cvs).
file.size.fatal check has been added in r14510.
This is fixed in 2.1.7 and 2.2_rc45.