This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Fortran Bug Fix Plan for 3.3.
- From: Toon Moene <toon at moene dot indiv dot nluug dot nl>
- To: gcc at gcc dot gnu dot org
- Date: Sun, 22 Sep 2002 15:45:01 +0200
- Subject: Fortran Bug Fix Plan for 3.3.
- Organization: Moene Computational Physics, Maartensdijk, The Netherlands
There are 14 non-closed bug reports for Fortran in GNATS (none for
libf2c), of which four are `suspended' and won't be fixed.
Here's how I plan to deal with them:
7681 ICE in compensate_edge, at reg-stack.c:2591
Most certainly a backend issue (and for ix86 at that).
Should I move it to `target' ?
7530 fortran statement works in 2.9.3 but not in 3.1.
I've set this one to `feedback' as it doesn't include
a self-contained test-case. I'll close it when it
becomes 2 months old.
7488 wrong line numbers after INCLUDE (resubmission).
A genuine frontend bug that I'll fix.
7388 Incorrect output with 0 based array of characters.
At first I thought this to be a bug resulting from the new code
to allow string(1:0) to pass array bound checking (new with
3.1), but it seems that the test case doesn't always fail (for
instance on sparc64) - so it could be a middle end issue.
7278 g77 "bug"; the executable misbehave.
The test case produces bogus output when compiled with -O2
-fno-automatic. I presume this is fixed by recent work from
Jan Hubicka fixing SPECfp2000's sixtrack, but I cannot check
this; I've put the bug report in `feedback' several weeks ago
but haven't got a reply yet.
7179 No error on DIMENSION A(250 000, 250 000) on 32-bit target.
Such an array is far too large to address on a 32-bit target
and should have been rejected by the frontend; will fix.
6742 Generation of C++ prototype for FORTRAN and extern "C".
Probably a documentation bug - either the documentation is
wrong, or it is too vague; will fix.
6545 GOT page/offset relocation out of range building LAPACK.
A problem on MIPS; rth says he can't reproduce - waiting for
feedback from David Billinghurst.
5900 Lapack regressions for g77 3.1
No current plan.
1832 list directed i/o overflow hangs, -fbounds-check doesn't
detect.
Known to be wrong, impossible to fix without changing run time
library API.
--
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
Join GNU Fortran 95: http://g95.sourceforge.net/ (under construction)