[fortran, patch] PR34402, accepts invalid init expr

[Tobias Burnus]

Hi Daniel,

Daniel Franke wrote:
> there's a condition on initialization expressions, that allocatable
components
> shall not be initialized with data but NULL(). Attached patch adds a
check for
> this. The testcase is the reporter's.

Thanks for the patch. Can you add an initialization expression which
uses NULL()? I only found it in
alloc_comp_{constructor,initializer}_1.f90, but those are not init expr.

> The wording of the error message is a bit awkward and longish, any other
> suggestions?

The patch uses:
"The data initialization in the derived type constructor at %L for the
ALLOCATABLE component '%s' is invalid in initialization expressions"

In standard lingo one does not use "derived-type constructor" but
"structure constructor", which is 3 letters shorter but I am not sure
whether it is clearer. I came up with the following version (and using
optionally the term "allocatable"):

"Invalid initialization expression for [allocatable] component '%s' in
structure constructor at %L"


> 2009-12-10  Daniel Franke  <franke.daniel@gmail.com>
>
>     PR fortran/34402
>     * expr.c (check_alloc_comp_init): New.
>     (check_init_expr): Verify that allocatable components are not
>     data-initalized.
>
> 2009-12-10  Daniel Franke  <franke.daniel@gmail.com>
>
>     PR fortran/34402
>     * gfortran.dg/alloc_comp_init_expr.f03: New.
>
>
> Regression tested on i686-pc-linux-gnu. Ok for trunk?

OK with the additional NULL() init expr and considering a shorter
version of the error message.

Tobias