On Tue, 2005-05-24 at 09:14 -0700, Shahms King wrote: > Thank you for doing this Jeremy. One minor nit to pick is that some of > these failed builds are impossible to debug and hence fix without more > information than is present in the <package>.failure.log file. The most > information that file provides is: "the build failed, if you could > access this local file, you might be able to figure out why. nya, nya, > nya". Which, when you're trying to figure out why the build failed on > an architecture to which you have no access, is more than useless and > quite a bit frustrating. The 4.5M rpm log is, of course, present when > the build succeeds. This seems, uh, backwards to me. Yeah, those are the ones that are harder... but most of the failures (luckily) don't fall into this category. So if we get the ones that are diagnosable, we'll be making good headway. > Seth? Any chance you could change those build scripts to copy the > rpm.log file when the build fails in addition to when it succeeds? >From looking at the code, it looks like that information should be getting written to the failed.log. Some of the ppc failures are absolutely bizarre, though. I'm working on setting up a machine locally to try to reproduce and figure out what's going on with some of them. Jeremy