On Tue, Jan 03, 2006 at 10:19:13AM -0500, Benjamin Mesing wrote: > Also note that it > was only one feature of packagesearch which broke (after all > packagesearch does only recommend xterm and works without it). But my > point was that such (unforseeable) things might break things in testing. Sure, so the right answer is to understand what happened, work out why it happened, and see what we can do to make sure it doesn't happen again. Even then, you'll never get zero breakage ever, but continually reducing the classes of breakage is a worthwhile activity in and of itself. > But I have taken the point of the replies. Second hand experience is > nothing I will base my judgement on (and make it public) any more. > > So please disregard my statement against "testing" I'd suggest a better approach is to take the second hand experience, try reproducing it in more concrete form, and work out what we can do to avoid it in future. This is all software, so doing it is pretty much just an exercise in working out exactly what we actually want to have happen. (In this case, xterm could have had a conflict with your package to avoid screwing users over unnoticed; and we could (in future) have added a note to the testing scripts to not allow upgrades of xterm until a fixed version of packagesearch is also included) The only problem is, we can't fix these problems if we only hear vague descriptions after the fact. Precise ones after the fact, or vague ones that are still current, otoh, we have a chance at. Cheers, aj
Attachment:
signature.asc
Description: Digital signature