This is the mail archive of the
mailing list for the GCC project.
Re: [Patch, fortran] PR29490 - ICE whilst building formal/actual interface
FX Coudert wrote:
Do a grep on GFC_TYPE_ARRAY_LBOUND and GFC_TYPE_ARRAY_UBOUND. You will
find that if the former is not set then the latter is not. However, if
this were not to be the case, it would make more sense to set LBOUND
from UBOUND, rather than to do an assert. I could put that branch in,
if you like.
In this case, trans-types.c(gfc_get_nodesc_array_type)
declines to set GFC_TYPE_ARRAY_LBOUND or GFC_TYPE_ARRAY_UBOUND
and gfc_set_interface_mapping_bounds. The solution is to use
the descriptor values for both bounds. The testcase is that
provided by the reporter.
I would just like confirmation that, if GFC_TYPE_ARRAY_LBOUND is not
set, then GFC_TYPE_ARRAY_UBOUND is garanteed not to be set either.
Perhaps an assert in the code would help?
BTW ironically, I was unable to find a call sequence that made use of
this array declaration; possibly it exists but I could not find a use
Thanks - I will commit on Saturday or Sunday, when I am back at base.
PS I finally got so sick and tired of my ISP's email web interface that
I decided to shift email to my portable permanently