I guess not directly with salome meca, because this package comes with its own prerequisites (2,3 GB)
The only clean way what I could think of would be to build our own Salome Meca by building Salome and Aster in the normal way with all prerequisites normally installed in ubuntu and then upgrade salome with the aster module.
the question is, if this would be worth the effort ...
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
actually you are completely right; I used the official binaries and repackaged them in a DEB to simplify deployement. However, the pre-requisite libraries makes it ultra big; but even though I tried several times, I never managed to build a full version of Salome with the Aster (and other) extra modules. The source codes for these additionnal modules are available but the compile/installation instruction are not sufficient to reproduce the Salome-Meca package (at least for me ;-)
The only solution I see to reduce the size of this package would be to delete some of the "pre requisite" libraries that exist and have the same version on the system (i.e de-duplicate the libraries). This could work, but I need to find a way to automate this tedious process.
If you are willing to contribute; you could try to spot the embedded "system" libraries and try to delete them to see if it would still work... I know that it worked on some previous versions of Salome-Meca.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I tried your suggestion. Salome works e.g. with removing the Python modules.
But then you can't import them anymore in the console. So Salome may work, but not without breaking anything.
I also tried to create symbolic links to the python interpreter of the system, instead of
the one of Salome, but without any success.
Rebuilding Salome Meca seems also not to be an option, because EDF build their version with the Commercial Intel Compiler. If we want to rebuild Salome Meca ourselves with an Intel Compiled Aster there may be license troubles waiting for us.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
thanks for trying and reporting! I tend to use the python console fairly often so I don't really want to take this part out...
On my side, I have tried another thing: I have deleted all static libraries (*.a) found in BOOST in the prerequisite folder. As the static libraries are only useful when compiling a code, I guess we don't need it. The interesting thing is that BOOST is huge, and the static libs take about 1Gb of space. So that's not a small saving.
I also deleted the doc in the MED folder as it is also quite big and not needed in most cases (except for code development).
To make it smaller, we could also delete all the test cases for Code-Aster 10.8 and 11.3 (astest folders) but these are really valuable examples when we need to figure out how some commands work in practice so I don't feel like taking them out if not really needed. What do you think?
I will rebuild the package without static libs and docs for MED and BOOST and we will see how much is saved in the end...
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
perhaps we also could then take out Code_Aster 11 and put it also in it's own package.
Then we could avoid the problem of leaving things out, by simple making them optional for the people who need it.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Instead of using multiple packages, for example graphical package GIMP uses "add-ons" in the ubuntu software center that can be checked to include or not... perhaps that can be used here as well.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I guess not directly with salome meca, because this package comes with its own prerequisites (2,3 GB)
The only clean way what I could think of would be to build our own Salome Meca by building Salome and Aster in the normal way with all prerequisites normally installed in ubuntu and then upgrade salome with the aster module.
the question is, if this would be worth the effort ...
Hello,
actually you are completely right; I used the official binaries and repackaged them in a DEB to simplify deployement. However, the pre-requisite libraries makes it ultra big; but even though I tried several times, I never managed to build a full version of Salome with the Aster (and other) extra modules. The source codes for these additionnal modules are available but the compile/installation instruction are not sufficient to reproduce the Salome-Meca package (at least for me ;-)
The only solution I see to reduce the size of this package would be to delete some of the "pre requisite" libraries that exist and have the same version on the system (i.e de-duplicate the libraries). This could work, but I need to find a way to automate this tedious process.
If you are willing to contribute; you could try to spot the embedded "system" libraries and try to delete them to see if it would still work... I know that it worked on some previous versions of Salome-Meca.
Hi!
I tried your suggestion. Salome works e.g. with removing the Python modules.
But then you can't import them anymore in the console. So Salome may work, but not without breaking anything.
I also tried to create symbolic links to the python interpreter of the system, instead of
the one of Salome, but without any success.
Rebuilding Salome Meca seems also not to be an option, because EDF build their version with the Commercial Intel Compiler. If we want to rebuild Salome Meca ourselves with an Intel Compiled Aster there may be license troubles waiting for us.
Hi,
thanks for trying and reporting! I tend to use the python console fairly often so I don't really want to take this part out...
On my side, I have tried another thing: I have deleted all static libraries (*.a) found in BOOST in the prerequisite folder. As the static libraries are only useful when compiling a code, I guess we don't need it. The interesting thing is that BOOST is huge, and the static libs take about 1Gb of space. So that's not a small saving.
I also deleted the doc in the MED folder as it is also quite big and not needed in most cases (except for code development).
To make it smaller, we could also delete all the test cases for Code-Aster 10.8 and 11.3 (astest folders) but these are really valuable examples when we need to figure out how some commands work in practice so I don't feel like taking them out if not really needed. What do you think?
I will rebuild the package without static libs and docs for MED and BOOST and we will see how much is saved in the end...
Ok that's a great Idea!
Perhaps a reasonable solution would be to break the big Package into smaller ones.
I.e. make Packages
salome-meca-main
salome-meca-boost
salome-meca-doc (examples)
perhaps we also could then take out Code_Aster 11 and put it also in it's own package.
Then we could avoid the problem of leaving things out, by simple making them optional for the people who need it.
Instead of using multiple packages, for example graphical package GIMP uses "add-ons" in the ubuntu software center that can be checked to include or not... perhaps that can be used here as well.