[Patch, fortran] PR29786 - [4.1/4.2/4.3 Regression] Initialization of overlapping variables: Not implemented

Paul Richard Thomas paul.richard.thomas@gmail.com
Fri May 25 04:15:00 GMT 2007


<on-list>

> somehow there are some Windows linebreaks in the file.
> patch chokes here because of a single "\r" on line 400. If I remove that
> one, patching works.
> (patch still writes a couple of times "(Stripping trailing CRs from
> patch.)".)

Bother - Tobi Schlueter picked me up for that a few weeks ago.
>

> I glanced at the changes and they seem to be ok, the gcc also bootstraps
> on x86_64-unknown-linux-gnu and the original testcase work. However, I
> get an ICE for the following program. g95 and ifort print "1 4". (NAG
> f95 prints garbage at runtime.)

Ahhh.  I more or less ignored derived types because of the legacy
aspect of this PR; apparently to the detriment of the patch.  I am
pretty sure that I know where the problem will be.

Thanks

Paul



More information about the Fortran mailing list