Re: new test failures

[Jim Meyering]

Akim Demaille wrote:
> Hi Jim!
>
> Le 17 févr. 2012 à 09:59, Jim Meyering a écrit :
>
>> This is a good reason to run each test in its own sandbox.
>
> Yes, definitely.
>
>> Have you looked at gnulib's init.sh framework?
>> It creates (and automatically removes upon termination)
>> a temporary directory for each test.  To use it, you add
>> a line like this at the top of your script:
>>
>>    . "${srcdir=.}/init.sh"; path_prepend_ ../src
>>
>> where in the above example, ../src specifies a directory
>> to prepend to your shell's search path, so you don't have to
>> use long absolute (risky, hard to reproduce) or often-unreadable
>> relative file names for programs you test.
>
> That might be an option, thanks for the pointer!
>
>
>> Oh!  I've just looked and see that bison is already there:
>>
>>    http://hydra.nixos.org/jobset/gnu/bison-master
>
> Great!!!  Thanks a lot for the pointer.
>
>> Now, all you need to do is request that a few names (or mailing list
>> address) be added to its config so that you are notified of state
>> changes.  The address@hidden mailing list is probably a good
>> place to start.
>
> I'll drop them a note, thanks!  Would bug-bison be appropriate
> in your opinion, or that would overflow subscribers?

You might want to try it personally for a month or so,
to see how much traffic it generates.  There are occasional
false-positives, e.g., when the builder runs out of resources.