Closed sakhan007 closed 9 years ago
I am unable to reproduce this. Could you please paste conda info -a
so we can see the environment variables too? In the meanwhile, please do conda update --all
in the environment.
After doing conda update --all
, the conda info -a
returns: http://pastebin.com/Bj38KQEh
I don't think this will solve the issue, but just to check: would you try this?
$ bash
$ unset LD_LIBRARY_PATH
$ python poisson_convergence1.py
In the meanwhile, more questions:
/usr/lib/gcc/x86_64-linux-gnu/4.8/libgfortran.so
present in your system?If you don't have libgfortran, then you will need it. There might be a way to provide it in conda, but I would have to make sure first.
If you already have it, then I still won't know what's happening and I will request you to do the following:
$ instant-clean
$ echo 'from dolfin import *
> f = Expression("5*pi*pi*sin(pi*x[0])*sin(2*pi*x[1])")
' > test.py
$ python test.py
and upload to pastebin all the files that appear under ~/.instant/
.
For the record: there's actually the option to install gfortran and libgfortran in conda: https://groups.google.com/a/continuum.io/d/msg/conda/HZlV2LXxWhs/UM_EgEeiYZIJ.
Here is the answer to every question I was asked: http://pastebin.com/0h7kN2NP
I don't think I need to install libgfortran, correct???
~.instant/
contained too many files, so here is the archive: https://www.sendspace.com/file/31ecqv
Actually, as a last step, try to remove libgfortran from conda (conda remove libgfortran
). If that doesn't work, I will have to inspect your files...
conda remove libgfortran
removed libgfortran as conda package.
Yes, I meant to try to rerun the script without the libgfortran
conda package.
The above archive is when I run possion_convergence1.py
and here is when I run test.py
: https://www.sendspace.com/file/myowfl after instant-clean
The conflict with libgfortran
dissapeared now, but apparently you mistyped the expression in the last case:
53: error: expected ‘)’ before ‘;’ token
values[0] = 5*pi*pi*sin(pi*x[0])*sin(2*pi*x[1];
Holy sh*t, now I found what's happening. You have an older version of gfortran than me! The program is looking for 4.8 and you have 4.6 (https://github.com/Juanlu001/fenics-recipes/issues/31#issuecomment-138124574). Can you confirm again what is your Linux distro and version? Now everything makes sense, but I'll have to figure out how to fix it.
Thanks for being so kind.
$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 12.04.5 LTS
Release: 12.04
Codename: precise
Updated archive after correcting the mistyping in test.py: https://www.sendspace.com/file/c9pp7i
Moreover my gcc version is 4.8.1 and gfortran is 4.6.4. I always thought that gfortran comes with gcc.
On 2015-09-07 01:16, sakhan007 wrote:
Moreover my gcc version is 4.8.1 and gfortran is 4.6.4. I always thought that gfortran comes with gcc.
Yeah, that looks like a strange mismatch for me too!
I assume that you cannot upgrade either gfortran or Ubuntu (FEniCS developers dropped Ubuntu 12.04 testing this year), so there are two options:
gfortran
and libgfortran
packages. Unfortunately I won't have time to do it until next week probably.I can upgrade gfortran and Ubuntu both as I have super user right. What do you suggest?
Sorry, I automatically assumed you were in a shared server or something!
Well, if you find a way to install gfortran 4.8 in your Ubuntu version that would be perfect. Obviously upgrading to a newer version of Ubuntu also would work but I understand that this can be a total mess.
Anyway, if you are not in a hurry and wait for the new package I would use your help to try if everything works :)
Ok I will wait.
I'm getting some errors while trying to build instant with anaconda gcc and gfortran:
https://groups.google.com/a/continuum.io/d/msg/anaconda/lmJatGVqklU/QVIfYFkWCAAJ
There are some suggestions to improve portability here:
https://github.com/conda/conda-recipes/pull/279#issuecomment-98254389
I would need to see how to fit them into instant. I might need to submit an upstream patch, but definitely I will need to look at it with some care. @sakhan007 You should probably update gfortran because I will need some weeks to get this right.
Does just updating gfortran would do the trick? or I have to build the whole fenics afterwards??
In principle updating gfortran should be enough. On Sep 14, 2015 10:16 AM, "sakhan007" notifications@github.com wrote:
Does just updating gfortran would do the trick? or I have to build the whole fenics afterwards??
— Reply to this email directly or view it on GitHub https://github.com/Juanlu001/fenics-recipes/issues/31#issuecomment-139993717 .
Hey guys, I didn't read the above thread too closely, but I noticed you referenced conda/conda-recipes#279.
Just FYI, for best results, you should use the latest version of the gcc
package, which is only available via the asmeurer
channel: conda install -c asmeurer gcc=4.8.5
. That version also includes some important fixes from conda/conda-recipes#401.
(BTW, the gcc
package includes gfortran
, and the libgcc
package includes libgfortran
.)
Theoretically, the conda guys will eventually push that gcc
package over to the main anaconda channel. Until then, you'll need to specifically reference the asmeuer
channel (or copy that package to your own channel, of course.) Just to repeat: Don't use the gcc 4.8.2
package that conda provides by default. It has portability problems.
Hi @stuarteberg, do you know if the Anaconda guys already pushed the latest fixes to the main channel? I'm waiting for that to fix my own packages. Thanks!
do you know if the Anaconda guys already pushed the latest fixes to the main channel?
It looks like they finally did: https://anaconda.org/anaconda/gcc/files
$ conda search gcc
Fetching package metadata: ....
gcc . 4.8.2 5 defaults
. 4.8.5 3 defaults
libgcc . 4.8.4 1 defaults
. 4.8.5 1 defaults
Thanks a lot for your answer, I will try to fix our issues now.
I pushed some changes that should fix these issues, but now I'm getting build errors related to PETSc. This is the neverending story :)
New packages uploaded. @e-aranda @sakhan007 Could you try updating DOLFIN to see if the errors are gone already? In principle the packages should be much more portable now.
Juan Luis,
thanks for your effort, but... I installed gcc-4.8 in my system and everything was fine. In order to try the new packages, I uninstalled gcc-4.8 and update dolfin from your channel. Then, running demo_poisson.py gives the following:
$ python demo_poisson.py
Vendor: Continuum Analytics, Inc.
Package: mkl
Message: trial mode expires in 30 days
Calling DOLFIN just-in-time (JIT) compiler, this may take some time.
Calling DOLFIN just-in-time (JIT) compiler, this may take some time.
Calling FFC just-in-time (JIT) compiler, this may take some time.
Calling FFC just-in-time (JIT) compiler, this may take some time.
Solving linear variational problem.
libGL error: unable to load driver: nouveau_dri.so
libGL error: driver pointer missing
libGL error: failed to load driver: nouveau
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 156 (GLX)
Minor opcode of failed request: 3 (X_GLXCreateContext)
Value in failed request: 0x0
Serial number of failed request: 58
Current serial number in output stream: 59
(fenics27)$
How can I undo the changes I made?
Ernesto
Hello @e-aranda, sorry for the inconvenience! Apparently I introduced some problem with the graphics which I have yet to understand. I will open a new issue, in the meanwhile you can manually install the previous build and run $ conda install /tmp/linux-64_dolfin-1.6.0-py27_2.tar.bz2
.
I tried
$ instant-clean
but it did not solve this problem.The compile.log, mentioned in the above error log is: http://pastebin.com/U7fhfEiX
The conda list: http://pastebin.com/43KDy6K3