Patch 2.5.9 tries to be "smart" and determine whether a file it is about to patch is under perforce revision control. However its heuristics are oversimplified and break when used with emerge/portage. The solution is to either change emerge to unset P4PORT, P4USER and P4CONFIG; or to modify patch to actually ask p4 whether the file in question is under control or not. The relevant section is in util.c but I'm not sure which approach to take. (I ran into this when trying to emerge openoffice-ximian after emerging perforce and having P4CONFIG=".p4config" set in my environment) Reproducible: Always Steps to Reproduce: 1. 2. 3.
epatch now unsets the P4* env vars
As a note: it seems that diffutils-3.3 doesn't have perforce-related code any longer.
(In reply to Michał Górny from comment #2) > As a note: it seems that diffutils-3.3 doesn't have perforce-related code > any longer. I'm sorry but I was wrong here. For some stupid reason, I assumed that patch was part of diffutils.
patch-2.7.6 still checks for these variables. Should we include them in the ENV_UNSET list for EAPI 7?
(In reply to Ulrich Müller from comment #4) > patch-2.7.6 still checks for these variables. Should we include them in the > ENV_UNSET list for EAPI 7? I don't see any negative impact on patching when I set P4CONFIG in the environment. Closing again.