[darcs-users] darcs patch: make unit's return value depend on all tests

Trent W. Buck trentbuck at gmail.com
Sat Jan 3 03:15:57 UTC 2009


Florent Becker <florent.becker at ens-lyon.org> writes:

> Fri Jan  2 19:49:30 CET 2009  Florent Becker <florent.becker at ens-lyon.org>
>   * make unit's return value depend on all tests
>
> Running the unit tests now returns something meaningful (the number of
> failed tests). On the other hand, these tests have been failing for
> quite a while… apparently since before we switched to quickCheck
> 2.1. Are we all ignoring them?

I think the "make unit" target generally has been ignored, and people
have only been running "make tests" which runs the functional tests.

Certainly *I* haven't run them because QuickCheck 2.1 isn't available
for Debian, so even though I have since capitulated and installed it via
cabal (ick), I got out of the habit of running it.

The code itself looks reasonable to me, just glancing at it.

The ./unit program already prints out (to stdout) which tests fail, so
to me the important thing is that this makes ./unit exit unsuccessfully
if there is a failed test (previously, it seems, it ALWAYS succeeded).

> Context:
>
> [Workaround issue1292: disable Haskeline by default, for now.
> Judah Jacobson <judah.jacobson at gmail.com>**20081230062721
>  Ignore-this: 36e64538118d4788a6f6d50a6bbdb930
>  The Haskeline backend doesn't return non-ASCII input in a way that Darcs can process.  The next
>  release of Haskeline should make it easier to resolve this issue.
> ]
> ...
> [TAG stable 2008-11-17 (with 2.1.2)
> Eric Kow <kowey at darcs.net>**20081117104434]
> ...
> [TAG 0.9.5
> droundy at abridgegame.org**Wed Apr  9 08:01:24 EDT 2003]

Why did Darcs generate such a huge context?!  Shouldn't it only go back
as far as the first tag shared with http://darcs.net/ (assuming that's
what you "darcs sent" against)?



More information about the darcs-users mailing list