New warnings for METAD adn WALKERS_MPI

[Iximiel]

Description

This should resolve #926:

METAD now throws if the user ask for WALKERS_MPI without a MPI installation and also if the MPI routines are not set up.

I wanted the changes testable so:

• I added a static variable to the Communicator to have the user access to the information "plumed is compiled with MPI" at runtime.
• I created a small non standard reg test to test the error thrown without using external tools and that tests exactly the needed messages
• I added the thrown exceptions with the messages set up to pass the new test

This is a draft because I wanted to show the solution that I have found, and I have not added the new test to the github workflow

Type of contribution

• changes to code or doc authored by PLUMED developers, or additions of code in the core or within the default modules
• changes to a module not authored by you
• new module contribution or edit of a module authored by you

Copyright

• I agree to transfer the copyright of the code I have written to the PLUMED developers or to the author of the code I am modifying.

• the module I added or modified contains a COPYRIGHT file with the correct license information. Code should be released under an open source license. I also used the command cd src && ./header.sh mymodulename in order to make sure the headers of the module are correct.

Tests

• I added a new regtest or modified an existing regtest to validate my changes.
• I verified that all regtests are passed successfully on GitHub Actions.

[GiovanniBussi]

Thanks @Iximiel ! A few comments:

• You can directly access the macro __PLUMED_HAS_MPI. Notice that any code including Communicator.h is currently forced to define all the relevant macros or it will not work.

• In general, I would avoid static member variables, since they can be modified by mistake with const_cast. I think using a function that returns a bool is more robust.

• An alternative way is to use directly Communicator::initialized() which also check if MPI is initialized. I think this is even better since you want to make sure that this is true

• Finally, in src we only keep source code and not tests. All tests are supposed to go in regtests, so that (a) they are executed on GitHub actions and (b) they are not installed

[GiovanniBussi]

Here I think you could just remove the first test.

Beside this, the way to raise the warnings is correct!!

[Iximiel]

In my first draft i had a single error message with "MPI is not initialized or plumed is not compiled with MPI", but I think is more clear for the user to have two more specialized messages, and easier to test for the developers

[GiovanniBussi]

I think you are right

[Iximiel]

Thanks @Iximiel ! A few comments:

* You can directly access the macro __PLUMED_HAS_MPI. Notice that any code including `Communicator.h` is currently forced to define all the relevant macros or it will not work.

Yes, but I set up the test as a completely external code, so when I compile the test I'm not seeing the compile-time definitions of PLUMED, or at least, I do not know how to access to them.

* In general, I would avoid static member variables, since they can be modified by mistake with const_cast. I think using a function that returns a bool is more robust.

My bad, now Communicator::PlumedHasMPI() is a function

* An alternative way is to use directly Communicator::initialized() which also check if MPI is initialized. I think this is even better since you want to make sure that this is true

when Communicator::initialized() returns false I don't know if it is because PLUMED is compiled without MPI or the MPI communication is not initialized

* Finally, in src we only keep source code and not tests. All tests are supposed to go in regtests, so that (a) they are executed on GitHub actions and (b) they are not installed

the test is in not in src/, but in unittest. Since is very far from the standard you use in the regtest directory, I though it was a good idea to keep it separated before the full merge

[carlocamilloni]

Hi @Iximiel, this is good with me! as a side note, the test that is failing now is a "style check" one. We enforce some basic styling in the code and you can get it fixed by running "make astyle" in src/ and commit again

[GiovanniBussi]

@Iximiel regarding this point:

Yes, but I set up the test as a completely external code, so when I compile the test I'm not seeing the compile-time definitions of PLUMED, or at least, I do not know how to access to them.

You are right. However, since we use some of the compile-time definitions of PLUMED in header files (actually very few, but __PLUMED_HAS_MPI is currently one of those), in order to properly compile source files including those header files you need to use the proper compile-time definitions. This is an (unsolved) problem for anyone willing to use plumed as an external library. Indeed, I am not aware of general solutions to this problem. A partial solution is that all those macros are included in the installed file lib/pkgconfig/plumedInternals.pc, but this is only for packages using pkgconfig. I guess for packages using cmake it should be straightforward.

Given this, once you bring your test within the regtest library, this is designed to make the macros available when compiling regtests.

So, I think it's fine using these macros in regtests. But it's also ok (and I think better) to write a small function that makes the bool available for outside usage.

[GiovanniBussi]

My bad, now Communicator::PlumedHasMPI() is a function

another stylistic convention that we have is: classes upper case, functions lower case. So perhaps it's more consistent to have Communicator::plumedHasMPI()

:-)

[codecov-commenter]

Codecov Report

Patch coverage: 100.00% and project coverage change: +0.09 🎉

Comparison is base (aadcb7e) 85.73% compared to head (1a83631) 85.82%.

❗ Current head 1a83631 differs from pull request most recent head 3f992bf. Consider uploading reports for the commit 3f992bf to get more accurate results

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #931      +/-   ##
==========================================
+ Coverage   85.73%   85.82%   +0.09%     
==========================================
  Files         600      600              
  Lines       53472    54722    +1250     
==========================================
+ Hits        45843    46964    +1121     
- Misses       7629     7758     +129     

Impacted Files	Coverage Δ
src/crystallization/EnvironmentSimilarity.cpp	94.06% <ø> (+0.02%)	⬆️
src/drr/DRR.h	100.00% <ø> (ø)
src/tools/Communicator.h	89.47% <ø> (ø)
src/tools/SwitchingFunction.cpp	95.95% <ø> (+0.14%)	⬆️
src/ves/TD_Multicanonical.cpp	96.53% <ø> (+0.10%)	⬆️
src/ves/TD_Uniform.cpp	98.11% <ø> (+0.11%)	⬆️
src/bias/MetaD.cpp	87.95% <100.00%> (+0.39%)	⬆️
src/drr/DRR.cpp	93.33% <100.00%> (+0.81%)	⬆️
src/drr/DynamicReferenceRestraining.cpp	90.55% <100.00%> (+0.29%)	⬆️
src/drr/drrtool.cpp	93.44% <100.00%> (-1.87%)	⬇️
... and 2 more

... and 262 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[Iximiel]

I have moved the new test in the regtest folder and I did some addition to make it work via make check in the main directory and with make in the regtest directory with an environment ready to launch an installed plumed, moreover is compatible with make reset

I runned astyle on the src, I think this can exit the draft state