On Tue, 2020-02-18 at 08:57 +0100, Thomas König wrote:
> Hi David
>
> in principle, any valid test case (especially for an ICE) should
> count as obvious and simple, so no approval should be needed.
>
> Having said that, I think I would prefer a copy of the original test
> case rather than an inc
Hi David
in principle, any valid test case (especially for an ICE) should count as
obvious and simple, so no approval should be needed.
Having said that, I think I would prefer a copy of the original test case
rather than an include statement. Although we usually do not change or remove
test c
PR analyzer/93774 reports an ICE in gfortran with -fanalyzer within
region_model::convert_byte_offset_to_array_index on a pointer of
incomplete type ("character(kind=1)[0:][1:0] * restrict").
This patch bulletproofs the routine against incomplete types, fixing
the ICE.
Successfully bootstrapped &
