Show exit codes in unified tests?

[Christian Ebert]

* Dirkjan Ochtman on Thursday, September 16, 2010 at 00:12:17 +0200
> On Wed, Sep 15, 2010 at 22:14, Matt Mackall <mpm at selenic.com> wrote:
>> This only show nonzero exit codes. Turning this on currently breaks 107
>> tests. And there are probably issues with different exit codes from
>> non-hg tools on different platforms we'd need to sort out too.
> 
> I like the idea. Can we come up with a method to alleviate
> cross-platform incompatibility? I.e. allowing specifying multiple
> allowed return codes.

Doesn't that defeat the purpose of testing for exit codes in the
first place? In case there are different exit codes, shouldn't
they be mapped to each platform? How big is the danger of false
successes? Or from the reverse angle: how much more precision
gives testing for a range of exit codes vs. simply testing for
success?

c
-- 
So is das Leben / Eben. / Eben und flach. / Ach.
     --->> http://www.blacktrash.org/hanullmann/