Re: [Patch, Fortran, OOP] PR 56261: seg fault call procedure pointer on polymorphic array

[Tobias Burnus <burnus at net-b dot de>]

Am 11.04.2013 23:38, schrieb Janus Weil:
> 2013/4/11 Tobias Burnus <burnus@net-b.de>:
> > Comment to Janus: It should be possible to disable the error check, 
> > e.g. by making it a warning. 
> > That's actually what the current code does: By default,
> > it only warns and does not print an error.
> No, at least the type mismatch and char-len checks always seem to
> throw an error ...

Only for function-result types:
  Error: Return type mismatch of function 'foo' at (1) (REAL(4)/COMPLEX(4))

But not for arguments:

subroutine foo(x)
  complex x(2)
end subroutine foo

real :: r(4)
call foo(r)
end

Warning: Type mismatch in argument 'x' at (1); passed REAL(4) to COMPLEX(4)


That's also how NAG's "f95 -dirty" handles it.

> > Thus, I think one should be strict about the requires-explicit-interface
> > diagnostic (= new code, using F90+), but for interface mismatch (= could be
> > old Fortran 66 code),  it should be either disabled or - as currently - just
> > be a warning.
> How about enabling it by default, but having -std=legacy disable the checks?

Hmm, I think I'd prefer to have for the arguments only a warning with 
-std=gnu - except for the function result value, which should always be 
an error. And also the "requires an explicit interface" checks should 
always use an error.

However, if one goes for an error with -std=gnu, the error should 
explicitly mention that "-std=legacy" disables the error. At least I do 
not recall all the special flags which enable certain features, nor 
whether such a flag exists at all. And I doubt that a normal user do.

(Actually, we already got bug reports about rejected code where the 
message explicitly stated after the error "This check can be disabled 
with the option -fno-range-check". Thus, some users do not even read the 
error message carefully.)

Tobias