[Patch, Fortran] PR33072 - Reject "use mod, only: operator(.foo.)" where "foo" is no operator

[Tobias Burnus]

:ADDPATCH fortran:

gfortran did not check whether operator(.symbol.) in USE statements were
really (user) operators or other symbols such as functions or
subroutines. This patch adds such a check.

Regression tested on x86-64/Linux. Ok for the Trunk?

<-----------Side note------------>
Unfortunately, this patch does not bring PR31298 any closer to being
fixed :-(

The problem is that I still don't fully understand how operators are
imported in modules.
Failing is currently:
- Multiple imports of operator(.useroperator.), example: see PR33072;
one needs to
do the same in load_operator_interfaces as in load_generic_interfaces,
but I don't
know how to handle the "i > 1" case.
- Renames of the kind "operator(+) => operator(.myplus.)"
<------End of Side note--------->

Tobias
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: module.diff
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20070816/a5797644/attachment.ksh>