Skip to content
Browse files

i386 builds are failing when they follow x86_64 builds.

This is VERY troubling, but for now just make sure they run first.
  • Loading branch information...
1 parent b431690 commit 82eba71f54942084555bca2e6689469b3a298f04 @erikogan erikogan committed Dec 2, 2010
Showing with 11 additions and 1 deletion.
  1. +11 −1 rpm/release/build.rb
View
12 rpm/release/build.rb
@@ -76,7 +76,17 @@ def limit_configs(configs, limits)
abort "ERROR: Cannot build '#{parts[2]}' packages on '#{rpmarch}'"
end
tree[parts.join '-']
- end.flatten
+ # All of a sudden the i386 mock builds are failing IFF they're run after an
+ # x86_64 build. I don't understand how separate processes in separate
+ # (chroot'ed!) environments can pollute each other, and I find it rather
+ # troubling. But for now the workaround is to sort by arch & do i386 first.
+ end.flatten.sort do |a,b|
+ ap = a.split(/-/)
+ ap.unshift(ap.pop)
+ bp = b.split(/-/)
+ bp.unshift(bp.pop)
+ ap <=> bp
+ end
end
def noisy_system(*args)

0 comments on commit 82eba71

Please sign in to comment.
Something went wrong with that request. Please try again.