bug#13832: Tests fail because make runs with more than one jobs

[Stefano Lattarini]

On 05/01/2013 02:17 AM, Joshua C. wrote:
> 2013/5/1 Stefano Lattarini <address@hidden>:
>> On 04/28/2013 09:39 AM, Joshua C. wrote:
>>> 2013/4/27 Stefano Lattarini <address@hidden>:
>>>>> Thanks for the promp solution! About the other part:
>>>>> What works for me is adding 'address@hidden@ --tag=CC', which I
>>>>> found on the internet. Can you reproduce this at your side?
>>>>>
>>>> Actually, I cannot even reproduce the failure at my side.  Which version
>>>> of Libtool are you using exactly?  Is that an upstream version, or has
>>>> been patched by your distro?
>>>>
>>>> Regards,
>>>>   Stefano
>>>
>>> I'm using the latest git available (commit:
>>> 68920ef8353f71c2e6c5c6b57a26cf55c0cff318) - unpatched with gcc-4.7.3
>>> on updated fedora17.
>>>
>> Thanks for the info.  I can indeed reproduce the bug using bleeding-edge
>> autoconf & libtool.  It is likely some kind of regression in one of those
>> tools.  Now I have a starting point to look better into the issue.  I hope
>> that won't take me other two months to do so ;-)
>>
>> Thanks,
>>   Stefano
> 
> Take your time. As I wrote, I found a solution that worked for me to
> pass the test (just a temporary solution).
> 
I've confirmed the failure is due to the bleeding-edge autoconf: things
still work fine using autoconf 2.69 and libtool 2.4.2, as well as using
autoconf 2.69 and the latest libtool from Git.  When I've proper time (I
hope soon), I'll try to use "git bisect" to pinpoint the exact commit
introducing the failure.  If anybody wants to beat me at it, though, he'd
be very welcome :-)

Regards,
  Stefano