This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: The complex problems
- To: Jeffrey A Law <law at cygnus dot com>
- Subject: Re: The complex problems
- From: Stephen L Moshier <moshier at mediaone dot net>
- Date: Wed, 14 Jul 1999 15:19:41 -0400 (EDT)
- cc: egcs at egcs dot cygnus dot com
- Reply-To: moshier at mediaone dot net
>This leaves us with a decision to make. Either we leave the default
>handling of complex values alone and document that it's totally busted
>for certain combinations of complex types and target systems, or we
>change the handling of complex so that we use -femulate-complex in
>those cases where the generic complex support in gcc is known to mess
>things up.
>
>Comments and suggestions are more than welcome at this juncture.
The simple fix that I use for float complex on alpha stimulated some
bugs that would sometimes delete the insn for the imaginary part.
They were fixed but they did not show up immediately. So I predict
that no matter how you fix it, something will break and cause a
regression. I think this should not be done as a last-minute hack.
Since complex has never worked on 64-bit targets it is not a
regression to leave it alone.