[Bug c++/115216] New: operator auto() gets instantiatied too eagerly

[Bug c++/115216] New: operator auto() gets instantiatied too eagerly

[olaf.krzikalla at dlr dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115216

            Bug ID: 115216
           Summary: operator auto() gets instantiatied too eagerly
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: olaf.krzikalla at dlr dot de
  Target Milestone: ---

Full example in:

https://godbolt.org/z/3hG5fh1PW

The example can be compiled either by commenting out line 24 or by swapping in
line 8 for line 9. In any case line 25 works. Thus there is a difference
between
the terms "xx[1]" and "xx.operator[](1)".

The error occurs with all recent gcc versions. It doesn't occur with any clang
version.

[Bug c++/115216] operator auto() gets instantiatied too eagerly

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115216

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 58281
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=58281&action=edit
testcase

Next time please attach the testcase or place it inline rather than just a link
to godbolt.

[Bug c++/115216] operator auto() gets instantiatied too eagerly

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115216

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
MSVC and ICC (EDG in GCC compat mode) also rejects the code for the same reason
as GCC. Are you sure this is not a clang issue?

[Bug c++/115216] operator auto() gets instantiatied too eagerly

[olaf.krzikalla at dlr dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115216

--- Comment #3 from Olaf Krzikalla <olaf.krzikalla at dlr dot de> ---
(In reply to Andrew Pinski from comment #2)
> MSVC and ICC (EDG in GCC compat mode) also rejects the code for the same
> reason as GCC. Are you sure this is not a clang issue?

To be honest I don't know. I have no idea, if a compiler is allowed to convert
a type before applying a operator, if there is already such an operator
available in the genuine class. Someone who knows the respective details in the
standard is required.