On Thu 2017-05-25 18:35:13 +0200, Chris Lamb wrote: > Daniel Kahn Gillmor wrote: > >> The failures on repeat are presumably because of something related to >> the gpg-agent daemon persisting across the tests. I'll look into >> invoking "gpgconf --kill all" at the start of the test suite > > Is that safe? What if I'm running gpgconf outside of the build? :)
I think i'll run it with the designated --homedir for the test suite itself, which should be safe. Also, i may decide to change the testing homedir to be randomly-generated, inside /tmp, rather than having it in the build tree. frustrating and annoying to have to deal with this, but this is what we get for relying on the constrained length of un_path in the unix domain socket struct. :/ --dkg