specfem3d nightly build has failed

[buildbot-princeton]

See build 115.

[luet]

It looks like buildbot didn't run the tests after @durochat commit yesterday. I'll fix that. In the meantime, I think the errors in this build are legitimate. It reports some unused variables. @durochat please let me fix buildbot before you open a new PR with specfem3d. I'll e-mail you when I am done. Thanks, David

[durochat]

Yes there is 4 unused variables, I fixed that now. Indeed, the buildbot not warned me. I did a git commit - git push, I wait your modifications to open a new PR.

[luet]

The problem came from a limit on the size of the patch. It was set at 100 MB and the patch resulting from this PR was 118 MB. I have now raised the limit to 1 GB.

To me that raises the question of what we put in the git repo. I thought we had decided to keep only code sources there but now it looks like we keep data as well, which will increase the size and therefore the download time for the repo. For example the file,

utils/EXTERNAL_CODES_coupled_with_SPECFEM3D/AxiSEM_for_SPECFEM3D/Parfiles_and_scripts_for_examples/VM_examples_OLDS/2.5D_alps/run_specfem3D/DATA/model_2D.txt 

is 14 MB, and the directory,

utils/EXTERNAL_CODES_coupled_with_SPECFEM3D/AxiSEM_for_SPECFEM3D/AxiSEM_modif_for_coupling_with_specfem/MANUAL/AXISEM_TUTORIAL

is 27 MB. It looks like there are a bunch of pictures in this directory in 3 different formats: .pict, .pdf and .png. The .pict files are all of the order of 3 MB.

[luet]

@durochat you can open a new PR to commit your fixes. Thanks, David

[QuLogic]

I don't really understand why we have a copy of AxiSEM when upstream is quite active and probably would have accepted any necessary changes.

[komatits]

Hi Clément,

Please remove all the useless big files such as the AxiSEM manual and the pictures it contains. We only need to keep the modified source code.

Thanks, Dimitri.

On 09/18/2015 07:34 PM, David Luet wrote:

The problem came from a limit on the size of the patch. It was set at 100 MB and the patch resulting from this PR was 118 MB. I have now raised the limit to 1 GB.

To me that raises the question of what we put in the git repo. I thought we had decided to keep only code sources there but now it looks like we keep data as well, which will increase the size and therefore the download time for the repo. For example the file,

utils/EXTERNAL_CODES_coupled_with_SPECFEM3D/AxiSEM_for_SPECFEM3D/Parfiles_and_scripts_for_examples/VM_examples_OLDS/2.5D_alps/run_specfem3D/DATA/model_2D.txt

is 14 MB, and the directory,

utils/EXTERNAL_CODES_coupled_with_SPECFEM3D/AxiSEM_for_SPECFEM3D/AxiSEM_modif_for_coupling_with_specfem/MANUAL/AXISEM_TUTORIAL

is 27 MB. It looks like there are a bunch of pictures in this directory in 3 different formats: .pict, .pdf and .png. The .pict files are all of the order of 3 MB.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141515094.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[komatits]

Hi,

Already answered in the last few months: that is a patched version, and the patch is specific (and unpublished); also, we do not want to follow their developments for now otherwise we will need to modify our changes and formats all the time.

A devel branch is a devel branch...

Cheers, Dimitri.

On 09/18/2015 09:26 PM, Elliott Sales de Andrade wrote:

I don't really understand why we have a copy of AxiSEM when upstream is quite active and probably would have accepted any necessary changes.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141544623.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[durochat]

Hi all,

Ok, thanks ! I opened a new PR to fix the unused variables. Tomorrow or Monday, I'll see in details all the files in Axisem, and other directories, that we can delete. And I will commit the changes.

Thank you very much. Best whishes, Clément.

2015-09-19 14:16 GMT+02:00 Dimitri Komatitsch notifications@github.com:

Hi,

Already answered in the last few months: that is a patched version, and the patch is specific (and unpublished); also, we do not want to follow their developments for now otherwise we will need to modify our changes and formats all the time.

A devel branch is a devel branch...

Cheers, Dimitri.

On 09/18/2015 09:26 PM, Elliott Sales de Andrade wrote:

I don't really understand why we have a copy of AxiSEM when upstream is quite active and probably would have accepted any necessary changes.

— Reply to this email directly or view it on GitHub < https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141544623 .

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141659913 .

[komatits]

Thanks!

Yes, please delete as much old stuff as (reasonably) possible.

Dimitri.

On 09/19/2015 07:12 PM, Clément Durochat wrote:

Hi all,

Ok, thanks ! I opened a new PR to fix the unused variables. Tomorrow or Monday, I'll see in details all the files in Axisem, and other directories, that we can delete. And I will commit the changes.

Thank you very much. Best whishes, Clément.

2015-09-19 14:16 GMT+02:00 Dimitri Komatitsch notifications@github.com:

Hi,

Already answered in the last few months: that is a patched version, and the patch is specific (and unpublished); also, we do not want to follow their developments for now otherwise we will need to modify our changes and formats all the time.

A devel branch is a devel branch...

Cheers, Dimitri.

On 09/18/2015 09:26 PM, Elliott Sales de Andrade wrote:

I don't really understand why we have a copy of AxiSEM when upstream is quite active and probably would have accepted any necessary changes.

— Reply to this email directly or view it on GitHub <

https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141544623

.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

— Reply to this email directly or view it on GitHub

https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141659913 .

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141689730.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[QuLogic]

that is a patched version, and the patch is specific (and unpublished);

It's in the public here, so in some sense, it is published.

also, we do not want to follow their developments for now otherwise we will need to modify our changes and formats all the time.

That is fine, but it is not really taking advantage of the distributed nature of git. Instead of copying everything, there should be a fork with the necessary patches. It does not need to be updated regularly, but when the time comes, it would be much easier to do since you're really just on a branch of the existing repo. Right now, it's much more likely this copy would fall out of sync and make updating all the more worse.

[komatits]

I agree with that; but as we mentioned a couple of years ago when we switched to Git many developers are not that familiar with Git and use it the old way (I mean the SVN way: update, add your changes, commit). Not optimal, but hard to change I guess. (that's also the way I use it :-(

Cheers, Dimitri.

On 09/19/2015 09:40 PM, Elliott Sales de Andrade wrote:

That is fine, but it is not really taking advantage of the distributed nature of git. Instead of copying everything, there should be a fork with the necessary patches. It does not need to be updated regularly, but when the time comes, it would be much easier to do since you're really just on a branch of the existing repo. Right now, it's /much/ more likely this copy would fall out of sync and make updating all the more worse.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141701090.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[QuLogic]

I recommend anyone who is not yet comfortable with git to try reading the software carpentry tutorial, or attending a lesson in person if it's available. It really makes life easier if git is used as git and not git-as-svn.

[komatits]

Good idea! Could you add that to the Wiki at https://github.com/geodynamics/specfem3d/wiki ?

Thanks, Dimitri.

On 09/20/2015 09:56 AM, Elliott Sales de Andrade wrote:

I recommend anyone who is not yet comfortable with git to try reading the software carpentry tutorial https://swcarpentry.github.io/git-novice/, or attending a lesson in person if it's available. It really makes life easier if git is used as git and not git-as-svn.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141756455.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[QuLogic]

OK, I've added it.

[komatits]

Thanks!

Dimitri.

On 09/20/2015 11:31 PM, Elliott Sales de Andrade wrote:

OK, I've added it.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141835996.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

[komatits]

Hi all,

We are now done cleaning the coupling and we have talked to Tarje Nissen, the AxiSEM modifications that Vadim has made are going to be put in the official AxiSEM source code and thus no local copy will be needed any more.

Qinya and Ping Tong will also soon merge their nice coupling with FK ( see http://komatitsch.free.fr/preprints/GRL_Ping_Tong_2014.pdf ).

Coupling with DSM is already merged (see http://komatitsch.free.fr/preprints/GJI1_Vadim_2013.pdf ).

Best, Dimitri.

On 09/18/2015 09:26 PM, Elliott Sales de Andrade wrote:

I don't really understand why we have a copy of AxiSEM when upstream is quite active and probably would have accepted any necessary changes.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/624#issuecomment-141544623.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr