[ase-users] Atoms.set_calculator() deprecated?

[Ask Hjorth Larsen]

Dear Jakob,

Am Mi., 20. Mai 2020 um 18:03 Uhr schrieb Jakob Schiøtz via ase-users
<ase-users at listserv.fysik.dtu.dk>:
>
> Hi all,
>
> What was the reason for deprecating Atoms.set_calculator() in favor of just setting the .calc attribute?  It is still a function that is being called.

Having only one is better than having both.  We regularly take some
flak for having multiple ways of doing the same thing.  So I suggested
it in an issue which got 5 thumbs up, which is the current high score.

>
> And more importantly: How the !€#% do I call that "function" (which is not a function, but a property) of the Atoms class, when I have a subclass that now has to overload the .calc property but needs to call the original not-quite-a-function.  With functions that is easy, but my Python-fu is insufficient to do this with properties.  I end up with infinite recursions no matter what I try.

Huh, that's kind of strange.  It isn't exactly user-friendly.  I
rummaged around and got to this:

from ase import Atoms

class MyAtoms(Atoms):
    @property
    def calc(self):
        print('fget')
        return Atoms.calc.fget(self)

    @calc.setter
    def calc(self, calc):
        print('fset')
        Atoms.calc.fset(self, calc)

atoms = MyAtoms()
atoms.calc = 'hello'
print(atoms.calc)

The ugly part being that you'd expect properties to be kind of
transparent, rather than requiring subclasses to *know* that they're
properties.

>
> Or would anybody object to partly reverting this change, un-deprecate Atoms.set_calculator() and letting Atoms.calc be a property calling the get/set functions which actually do the work?

I think it might be okay to have getters/setters to help subclasses,
but I prefer to still deprecate the old API.  As long as remains clear
to users what the purpose of each thing is.

A real issue is that it's not clear what the subclassing rules are for
Atoms.  Subclassing complex objects is always finicky, having a
tendency to break abstractions that otherwise work.  So I always
prefer delegation.

Hmmm ...

Best regards
Ask

>
> Best regards
>
> Jakob
>
> --
> Jakob Schiøtz, professor, Ph.D.
> Department of Physics
> Technical University of Denmark
> DK-2800 Kongens Lyngby, Denmark
> http://www.fysik.dtu.dk/~schiotz/
>
>
>
>
> _______________________________________________
> ase-users mailing list
> ase-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users