dorsal team mailing list archive
-
dorsal team
-
Mailing list archive
-
Message #00092
Re: SLEPc problems in Mac OS X
On 8/18/10 6:32 PM, Anders Logg wrote:
> On Wed, Aug 18, 2010 at 06:26:22PM +0530, Harish Narayanan wrote:
>> On 8/18/10 6:16 PM, Anders Logg wrote:
>>> On Wed, Aug 18, 2010 at 01:44:26PM +0100, Garth N. Wells wrote:
>>>> On Wed, 2010-08-18 at 14:42 +0200, Anders Logg wrote:
>>>>> On Wed, Aug 18, 2010 at 01:46:18PM +0200, Anders Logg wrote:
>>>>>> On Wed, Aug 18, 2010 at 12:43:23PM +0100, Garth N. Wells wrote:
>>>>>>> On Wed, 2010-08-18 at 13:38 +0200, Anders Logg wrote:
>>>>>>>> I'm getting the following error message when installing with Dorsal on
>>>>>>>> my Snow Leopard MacBook:
>>>>>>>>
>>>>>>>> Unpacking slepc-3.1-p1
>>>>>>>> Building slepc-3.1-p1
>>>>>>>> Checking environment...
>>>>>>>> Checking PETSc installation...
>>>>>>>> Checking LAPACK library...
>>>>>>>>
>>>>>>>> ================================================================================
>>>>>>>> SLEPc Configuration
>>>>>>>> ================================================================================
>>>>>>>>
>>>>>>>> SLEPc source directory:
>>>>>>>> /Users/logg/Work/FEniCS/src/slepc-3.1-p1
>>>>>>>> SLEPc install directory:
>>>>>>>> /Users/logg/Work/FEniCS
>>>>>>>> PETSc directory:
>>>>>>>> /Users/logg/Work/FEniCS
>>>>>>>> Architecture "unknown" with double precision real numbers
>>>>>>>>
>>>>>>>> **
>>>>>>>> ** Before running "make" your PETSC_ARCH must be specified with:
>>>>>>>> ** ** setenv PETSC_ARCH unknown (csh/tcsh)
>>>>>>>> ** ** PETSC_ARCH=unknown ; export PETSC_ARCH (sh/bash)
>>>>>>>> **
>>>>>>>>
>>>>>>>> makefile:30: darwin10.0.0-cxx-opt/conf/slepcvariables: No such file or
>>>>>>>> directory
>>>>>>>> /Users/logg/Work/FEniCS/src/slepc-3.1-p1/conf/slepc_common_variables:2:
>>>>>>>> /Users/logg/Work/FEniCS/src/slepc-3.1-p1/darwin10.0.0-cxx-opt/conf/slepcvariables:
>>>>>>>> No such file or directory
>>>>>>>> /Users/logg/Work/FEniCS/src/slepc-3.1-p1/conf/slepc_common_rules:78:
>>>>>>>> /Users/logg/Work/FEniCS/src/slepc-3.1-p1/darwin10.0.0-cxx-opt/conf/slepcrules:
>>>>>>>> No such file or directory
>>>>>>>> make: *** No rule to make target
>>>>>>>> `/Users/logg/Work/FEniCS/src/slepc-3.1-p1/darwin10.0.0-cxx-opt/conf/slepcrules'.
>>>>>>>> Stop.
>>>>>>>> Failure with exit status: 2
>>>>>>>> Exit message: There was a problem building slepc-3.1-p1.
>>>>>>>>
>>>>>>>> Any ideas what goes wrong?
>>>>>>>>
>>>>>>>
>>>>>>> I thought that I fixed this yesterday. See if you can figure out what's
>>>>>>> going on in
>>>>>>>
>>>>>>> FEniCS/packages/slepc.package
>>>>>>>
>>>>>>> Garth
>>>>>>
>>>>>> ok.
>>>>>
>>>>> The problem seems to be PETSC_ARCH_TEMP. The way I understand this is
>>>>> that this should be set in package_specific_register as part of
>>>>> petsc.package and then propagated to slepc.package, but that variable
>>>>> seems to be empty (in both petsc.package and slepc.package).
>>>>>
>>>>> Where is PETSC_ARCH set?
>>>>>
>>>>
>>>> In the foo.platform file.
>>>>
>>>> It should really set wherever the decision on the PETSc build options is
>>>> made.
>>>
>>> Aha. Then your fix yesterday seems to be the problem. It overwrites
>>> PETSC_ARCH with PETSC_TEMP which is empty. It should check if
>>> PETSC_TEMP is non-empty an only then overwrite PETSC_ARCH. I'll see if
>>> I can fix it.
>>
>> :_(
>>
>> Harish
>
> This wasn't the problem... The PETSC_ARCH variable seems to
> be set correctly after all. It still doesn't work for me. Does it work
> for you Harish?
I have not pulled since the move to petsc/slepc 3.1. I have tried to do
the move more than once (before Garth's recent changes) and reverted
because of a lot of silly things like this which I wasn't keen on
looking deep into.
> Something that I think is strange is that PETSC_ARCH is not set until
> after the configure phase for SLEPc. Then I get the above error. If I
> move it up so it is set before configure, then SLEPc complains that
> PETSc has not been built for darwin10.0.0-cxx-opt.
I guess I will pull a fresh copy and start trying things. It is going to
take a while on my third world bandwidth.
Harish
References