------- Comment #29 from rguenther at suse dot de 2007-10-10 15:47 -------
Subject: Re: wrong types in character array/scalar binop
On Wed, 10 Oct 2007, pault at gcc dot gnu dot org wrote:
> ------- Comment #28 from pault at gcc dot gnu dot org 2007-10-10 15:44
> -------
> The patch below fixes the lot. It was not necessary in the end to touch
> trans-intrinsic.c. Once the appropriate, offending bit of trans-array.c was
> fixed, all the casting occurred correctly. The fixes to iresolve.c deal with
> the various ICEing testcases in the comments below and are partially based on
> FX's input.
>
> This now compiles an runs correctly.
>
> character(len=1) :: string = "z"
> integer :: i(1) = (/100/)
> print *, Up("abc")
> print *, transfer(((transfer(string,"x",1))), "x",1)
> print *, transfer(char(i), "x")
> print *, Upper ("abcdefg")
> contains
> Character (len=20) Function Up (string)
> Character(len=*) string
> character(1) :: chr
> Up = transfer(achar(iachar(transfer(string,chr,1))), "x")
> return
> end function Up
> Character (len=20) Function Upper (string)
> Character(len=*) string
> Upper = &
> transfer(merge(transfer(string,"x",len(string)), &
> string, .true.), "x")
> return
> end function Upper
> end
>
> and the code in achar, which Richard flagged up has become:
>
> char char.6;
>
> char.6 = (*(char[0:][1:1] *) atmp.3.data)[S.5][1]{lb: 1 sz: 1};
> (*(char[0:][1:1] *) atmp.4.data)[S.5][1]{lb: 1 sz: 1} = char.6;
nice.
> The patch even regtests but I will check tonto and cp2k before submitting.
Thanks for the efforts!
Richard.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31608
