Jonathan Tan <jonathanta...@google.com> writes: > On Thu, 3 May 2018 11:58:59 -0700 > Stefan Beller <sbel...@google.com> wrote: > >> > + test_must_fail git -C server serve --stateless-rpc <in >/dev/null >> > 2>err && >> >> Minor nit: >> Why do we pipe stdout to /dev/null ? > > Usually there's a binary packfile produced, but not in this case. I'll > remove it.
Hmm, when somebody breaks "git server serve", we probably would not want to see the binary spewed to the output while debugging it. So I'd probably keep the redirection---it may be an improvement to use ">out" and then checking it is empty after the expected failure.
