On 9/5/06, b.n. <[EMAIL PROTECTED]> wrote:
Absolutely right. But at this point shouldn't the non-tested package be moved to ~arch? Or at the very beginning of major GCC upgrade processes (like 3.3-->3.4 or 3.4-->4.1), should there be some automatic advice that "the following packages on your world have NOT been tested with the GCC/glibc/kernel/whatever version you're trying to switch to, are you sure to switch?"
The problem is there is no such database of "this was tested and works/doesn't work". The closest is bugzilla, with the blocking bugs on the gcc stabilization bug. But that only tells you what was tested and found to be broken, not what was tested and found to work. I don't think moving the known-broken packages to ~arch would be a good idea, as someone could just decide to stick with the previous version of gcc. Moving the packages to ~arch would force such users to unmask the package to keep using it, even if they do not upgrade gcc. However, for the next gcc-upgrade cycle, I plan to ask (as a userrep) that the gcc compilation bugs not be closed until the fixes actually make it to stable. That way at least the stabilization bug will continue to reflect just how broken the tree might get if it is pushed through. -Richard -- gentoo-user@gentoo.org mailing list
