Closed jhenin closed 3 months ago
end_of_step()
is a member of FixColvars
, which does not inherit from colvarproxy
and therefore does not override it. Am I missing something?
Sorry, my mistake.
NP. This is reminding us, though, that are testing in CI for hidden functions within the library, but not in the backends' proxies, where there is a ton of inheritance. If Clang is at the point where it can detect this kind of errors, it may be worth to switch building the backends with Clang rather than GCC.
(At least GROMACS and LAMMPS, where we know that these compilers are already tested in their native CI?).
The LAMMPS proxy is the only one to override
colvarproxy::end_of_step()
. It should probably call the base class method. Alternately, maybe the current content of end_of_step() is totally generic, and we don't want it overridden by any proxy: then it should be its own non-virtual function.