Repeat count for modern-style initializations

[certik]

@FortranFan

@Machalot

with

real :: var = 0.0

That is unlikely a “common suggestions for modernization”.

What might be FORTRAN code as

     REAL VAR
     DATA VAR /0.0/

will require analysis to ascertain whether the “implied SAVE” is really necessary. The refactoring decision will depend on this analysis.

More often than not, the modern Fortran replacement will be

   real, parameter :: VAR = 0.0

In a few cases, it might call for an object definition in the executable section of the code separate from the variable declaration:

   ..
   real :: var
   ..
   var = 0.0

In a few unfortunate situations, the modern Fortran replacement may need to

real, save :: var = 0.0 !<-- inform the reader explicitly the object has

Excellent, we ended up with the following warning in LFortran after a few iterations:

warning: Assuming implicit save attribute for variable declaration
 --> a.f90:2:9
  |
2 | real :: var = 0.0
  |         ^^^^^^^^^ help: add explicit save attribute or parameter attribute or initialize in a separate statement

Which enumerates the same three cases that you mentioned. So this is good.

[RonShepard]

@certik

[...] or initialize in a separate statement

Perhaps the term “initialize” is a little confusing here. In fortran, “initialize” usually means at compile/link time, not at run time. The third alternative in this situation is a run time assignment statement. That is “initialization” in a mathematical or algorithm context, but not in a fortran context.

[certik]

@RonShepard thanks for feedback. How would you formulate it, let’s say: “help: add explicit save attribute or parameter attribute or use a separate runtime assignment statement”?

[Harper]

Last time I put SAVE among the specification statements of a subroutine to save everything savable that was declared there, lfortran gave me a lot of warnings I had hoped it wouldn’t, and I had to put things like real,save:: in all the declarations of variables.

Is there a chance that one day we could use something like NO_SAVE where we can now use SAVE if we wanted that effect?

[Machalot]

@Harper

Is there a chance that one day we could use something like NO_SAVE where we can now use SAVE if we wanted that effect?

I’ve always wondered why there’s no nosave to disable implicit save.

[certik]

@Harper

Last time I put SAVE among the specification statements of a subroutine to save everything savable that was declared there, lfortran gave me a lot of warnings I had hoped it wouldn’t, and I had to put things like real,save:: in all the declarations of variables.

That’s a bug, I just reported it: Implement standalone `save` declaration line · Issue #4544 · lfortran/lfortran · GitHub, it should behave the same no matter where you put the explicit save. Update (12h later): and the bug is fixed. @Harper if you discover any other issue related to this, let us know.