[darcs-users] darcs patch: Accept issue1401: bug in get_extra commuting patches.

Trent W. Buck trentbuck at gmail.com
Mon Mar 30 09:04:58 UTC 2009


On Mon, Mar 30, 2009 at 10:26:36AM +0200, Petr Rockai wrote:
> "Trent W. Buck" <trentbuck at gmail.com> writes:
> 
> > I only get that for the legacy "hashed" and "old-fashioned-inventory"
> > types.  I believe the bug is not relevant to those repository formats,
> > but ICBW.
> >
> > For "darcs-2" format, you should get
> Ah, right, OK. However, it would make sense to still skip the test for
> non-darcs-2 runs (otherwise, it'd start failing spuriously once the bug is
> fixed).

> Something along the lines grep -q darcs-2
> $DARCS_TESTING_HOME/.darcs/defaults

Fugly.  That will break when anyone changes the way prefs are set up
for the tests.  But I'll do it for now...

> || exit 200 could possibly work (assuming exit 200 makes the harness
> skip the test... if there's no such thing, it might be worth
> patching the harness... otherwise you can make it just exit 0 and
> the test will "pass" automatically for non-darcs-2 repos).

Sorry, the closest you can get in the real world (AFAIK) is "exit 128"
and hope you're on a system where it wraps around.  But generally 0 is
the only successful condition.  Also, exit codes 126 and 127 are
typically reserved for "program not found", so you'd have to use, say,
"exit 125" and treat that exit code specially in ShellHarness.hs.

> Btw. I think there might be other instances of format-specific
> tests. Look around the testsuite maybe? Thanks!

I don't wanna.


More information about the darcs-users mailing list