On Wed, May 23, 2012 at 01:48:12PM +0200, Martin Costabel wrote:
> On 23/05/12 00:19, Alexander Hansen wrote:
>> On 5/22/12 2:46 PM, Hanspeter Niederstrasser wrote:
> []
>>> I just tried the following test on 10.7/xcode4.1, all with "fink -m
>>> --build-as-nobody rebuild openmpi":
>>>
>>> no Fink make installed: works
>>> make-3.81-2 installed: lock removal failure
>>> make-3.82-1 installed: works
>
>> Interesting...I tried again with 10.7.4/Xcode 4.3.2 and no Fink make and
>> the build worked for me. I've _never_ had a Fink make on this machine,
>> however, so the prior failure was also with Apple's make.
>
> This thing is not deterministic, as can be seen from the differing
> places where it hangs, when it hangs. I did some tests and tried varying
> some parameters to see whether it is completely random, and I think now
> it is. In my case (mac mini 4 core), the decisive parameters are
> certainly not build-as-nobody (which I have not been using), nor Fink's
> make (my first failing attempts were with make-3.82-1 installed, but it
> just now also hung with Fink's make removed).
>
> I remember now a related discussion of some 6 months ago on this list in
> the thread "Build of cyrus-sasl2-2.1.23-4 on Tiger/PPC fails due to some
> lock glibtool is waiting to be removed". There the problem was finally
> blamed on Fink's libtool, which tries to be helpful and protect parallel
> builds against a non-existing problem, the remedy killing the healthy
> patient.
>
> In that thread there were some deep contributions by pogma indicating
> that this might be solved in some recent version of libtool, but
> apparently nothing came of it, and the result is that cyrus-sasl2 now
> builds with "UseMaxBuildJobs: False". For openmpi, however, to forbid
> parallel builds would be, let's say, not quite optimal :-)
>
> The real fix is to not let libtool play its dumb locking game. What
> should work (I have no time now to test it) is removing the following
> lines from the various openmpi configure scripts:
>
> case $host_os in
> darwin*)
> lock_old_archive_extraction=yes ;;
> *)
> lock_old_archive_extraction=no ;;
> esac
>
> and replacing them simply with
>
> lock_old_archive_extraction=no
>
> --
> Martin
Martin,
You're welcome to make the change to the openmpi packages. As I can't
reproduce the failure, you are in a better position to test if the fix is
in place before committing.
Jack
>
>
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Fink-users mailing list
Fink-users@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.macosx.fink.user
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-users
