Gwern Branwen <[email protected]> writes:

> On Fri, Aug 7, 2009 at 7:33 AM, Petr Rockai<[email protected]> wrote:
>> I don't particularly care -- if you want, re-record it in small chunks. 
>> However
>> please note that the suggested sequence does not work (between 4 and 5, you 
>> get
>> a failure in the hlint script, unless you put it somewhere else than
>> tests/... between 6 and 7, haskell_policy breaks).
>>
>> Yours,
>>   Petr.
>
> Well, OK. I've been recording, but so far I haven't made the final change,
> since doesn't hlint still error out on the type witness stuff? That would
> break the test no matter what ignores and whatnot we put in the hints.

Please, check your facts. No, parse errors are not fatal for hlint -- it will
ignore the file and move on. It does not produce an error exit code either. And
finally, majority of the parse errors are just missing CPP support. See the
issue I have linked last time:
http://code.google.com/p/ndmitchell/issues/detail?id=137

Your patchset of course still breaks, because it doesn't do ratification. Maybe
it would be more productive to just use the working patch I sent? It only makes
sense to break up into small patches if you send a bundle that actually works
and does everything that needs to be done.

Also, your comments in the hint file are misleading. It is *not* the purpose of
the hint file to give style warnings. It is about *testsuite*. Fail/pass.
Explain failures (and only failures, not a zillion of warnings that are
unrelated to the failure). If you want to have style warnings, these belong to
post-build hook in Setup.lhs, or a separate script -- it doesn't make any sense
to have these in the testsuite.

Yours,
   Petr.
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to