Update actions in workflow and improve parallelisation and cache handling

[leonardocarreras]

Changelog

In general, the versions of the actions were updated for checkout, cache, download and upload artifacts. Aditionally, the parallel build with a static value in an environment variable -j2 was changed to the --parallel $(nproc) argument to the command to pick automatically the value.

In particular, other changes were done:

• build_test_linux_fedora_minimal.yaml
  • the build cache is updated with a new name to preserve for the same reference (branch, push) and not for the same commit
• build_test_linux_fedora.yaml
  • setup a build cache for dpsimpy & examples
  • archive in a different cache (not overwriting the others) per commit sha
  • changes to the archiving of build of cxx examples
  • input from the previous stage and recompile and save for the current compilation
  • jupyter notebook tests were parallelized
  • the package pytest-xdist is installed
  • cache is changed to separate per commit sha
  • comparison of results
  • previous cached result is used
  • test of binaries
  • the output of the cxx compilation is used as cache
  • villas examples
  • the needs is changed to examples (dpsimpy probably worked due old overwrites in the cache, when chaining the compilation outputs it started failing)
• build_test_linux_rocky.yaml
  • added cache to the clang compilation
  • added cache to the profiling compilation
• build_test_windows.yaml
  • added cache to the build directory per reference
• documentation-fein.yaml
  • changed node to v20
  • updated hugo and gh-pages actions
• run_and_profile_example.yaml
  • changed the container to sogno/dpsim.dev-rocky (we were still on centos)
• run_villas_example.yaml
  • changed the cache to the cxx examples build

[leonardocarreras]

I'm not sure we want caching for the profiling actions, because the point of this action was to have performance figures for each commit. However, we don't really use them, so I would also be fine with removing them.

I think if we do not plan to use them it would be okay to delete the cache or the action itself. I listen to opinions.

[n-eiling]

OK thanks for the additional research. Looks like our runners get 4 CPUs according to the github documentation. I don't need the profiling anymore, so I would be fine with removing them to make the CI run faster.

[leonardocarreras]

There is now a villas problem, I am unsure what might have been changed in villas in the last days:

In file included from /__w/dpsim/dpsim/dpsim-villas/src/InterfaceWorkerVillas.cpp:15:
/__w/dpsim/dpsim/dpsim-villas/src/../include/dpsim-villas/InterfaceWorkerVillas.h:8:10: fatal error: villas/exceptions.hpp: No such file or directory
    8 | #include <villas/exceptions.hpp>
      |          ^~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from /__w/dpsim/dpsim/dpsim-villas/src/InterfaceVillas.cpp:3:
/__w/dpsim/dpsim/dpsim-villas/src/../include/dpsim-villas/InterfaceVillas.h:8:10: fatal error: villas/exceptions.hpp: No such file or directory
    8 | #include <villas/exceptions.hpp>
      |          ^~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.

https://github.com/sogno-platform/dpsim/actions/runs/9402263953/job/25901074854#step:7:429

My view is that the error is not caused by this PR, as I can see problems in the pipeline for others...

[leonardocarreras]

Hi @n-eiling! I see that you are working on #299 and the updated container with a new villas version, thanks for this... Do you think that it will be fixing the new errors that we get? In your opinion could this PR go first? Should I retrigger the workflow of container.yaml with the older container to make the workflow to pass again?

[n-eiling]

This is due to an error in VILLAS. I already fixed it. Once it's merged the container should work again. Sorry for breaking this.

[leonardocarreras]

@n-eiling thank you for fixing, maybe with luck the merge does not take long :)

[m-mirz]

@leonardocarreras instead of removing the profiling we could also trigger this manually. In general, I think it is valuable to have the profiling in such a convenient way.

[sonarcloud[bot]]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
No data about Coverage
No data about Duplication

See analysis details on SonarCloud

[leonardocarreras]

@m-mirz Now it works as suggested (on-demand)