if i have a fix for the nailgun connection error we're seeing in twitter ci, and i can verify the bug exists in 1.8.x (because we saw it there too), would that be cause for a 1.8.x backport? it's very difficult to trigger without e.g. a 7000 file diff so i'm leaning towards no, but also, i doubt it would introduce other breakages (the fix is very small, i'm going to actually try to add testing now). it doesn't seem too hard to just fix it in 1.9.x since that release cycle is still open and just wait for anyone to make an issue about it for 1.8.x -- but i don't know if in general for these things we err on the side of "fix known bugs when feasible" (in this case it crashes pants, so that's bad, but it's also very difficult to repro even if you're trying to) vs "only introduce code into backports that's definitely necessary"