Eric Wong <[email protected]> wrote:
> I got one mysterious test failure here, once, and can't seem
> to reproduce it...
nearly 2 years later: I've hit it again, but once again cannot
reproduce it...
> +++ b/t/lei-p2q.t
> @@ -7,7 +7,7 @@ require_mods(qw(json DBD::SQLite Search::Xapian));
>
> test_lei(sub {
> ok(!lei(qw(p2q this-better-cause-format-patch-to-fail)),
> - 'p2q fails on bogus arg');
> + 'p2q fails on bogus arg') or diag $lei_err;
> like($lei_err, qr/format-patch.*failed/, 'notes format-patch failure');
$? does get printed (32768, so (>> 8) means 128), as it should be.
So I wonder if there's a place we drop the socket prematurely or
something else is amiss...
On a side note, lei's internal IPC could probably be simplified
a bit w/o sacrificing parallelism.
