On Wed, Mar 26, 2008 at 10:55:21AM -0700, Junio C Hamano wrote: > In the longer run, we probably would want to do something similar for all > tests to ensure a stable testing environment, don't we? $HOME/.gitconfig > may affect the way tested programs behave otherwise. We used to, as per a patch from Gerrit long ago: 2eb10ac... Set $HOME for selftests but it was removed during Dscho's config fix: 8565d2d... Make tests independent of global config files which used GIT_CONFIG instead. But that had some problems, which I fixed in: 8bfa6bd... fix config reading in tests and that patch explicitly suppresses lookup in the ~/.gitconfig and /etc/gitconfig files (implemented by ab88c363). So I think all is correct with .gitconfig files. However, I still think setting HOME is a good idea, because it eliminates one more variable in test runs. In particular, I know I have gotten cvsps into a funny state that was resolved by removing the ~/.cvsps cache. So it's probably worth doing as a preventative measure, even if it doesn't fix a specific bug. -Peff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
