This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
g77 regression fix patch for 1.1 branch
- To: law at cygnus dot com
- Subject: g77 regression fix patch for 1.1 branch
- From: Craig Burley <burley at gnu dot org>
- Date: Sat, 5 Sep 1998 23:50:57 -0400 (EDT)
- Cc: egcs-patches at cygnus dot com
- Cc: burley at gnu dot org
Jeff, could you install this in the 1.1 branch? More info is in
the news.texi patchlet.
Note that this is a minimal version of Toon Moene's recent patch.
Looking at the code his patch changes, I see even more fixups that
are probably worthwhile for the mainline. I've lost track of whether
his patch was applied to the mainline -- I'd rather it not be, since,
although it produces a more "elegant" libg2c, it does so by being
"more different" than netlib's libf2c than necessary. And the fact
that g77's libg2c is already different has caused bugs like the one
being fixed here.
So, I'm going to try to make some time in the next few months to
better reconcile libf2c and libg2c, by sending pertinent patches
(and some test cases, e.g. for complex-operation overlap bugs I
think libf2c has that libg2c doesn't) to dmg, etc. (dmg is the
maintainer of netlib libf2c.)
tq vm, (burley)
P.S. I've looked at the change to libf2c that triggered this bug
(which happened on the g77, not netlib, side, which is why this
is a regression, not a longstanding bug). I can't figure out why
e_wdfe was moved from dfe.c to sfe.c either. I'm going to ask
dmg about this. (Of course, if he moves it back, we'd better do
a better jobbing of moving *our* modified version over than before. :)
Sat Sep 5 23:06:56 1998 Craig Burley <burley@gnu.org>
* libI77/dfe.c (e_rdfe): Restore resetting of f__init
before returning, to avoid spurious recursive-I/O
diagnostic.
*** g77-e/gcc/f/news.texi.~1~ Fri Sep 4 20:21:42 1998
--- g77-e/gcc/f/news.texi Sat Sep 5 23:14:09 1998
***************
*** 6,10 ****
@c in the G77 distribution, as well as in the G77 manual.
! @c 1998-09-04
@ifclear NEWSONLY
--- 6,10 ----
@c in the G77 distribution, as well as in the G77 manual.
! @c 1998-09-05
@ifclear NEWSONLY
*************** though this can make getting a complete
*** 58,61 ****
--- 58,71 ----
of what a particular @code{egcs} version contains
somewhat more difficult.
+
+ @heading In 0.5.24 and @code{egcs} 1.1.1 (versus 0.5.23 and 1.1):
+ @itemize @bullet
+ @item
+ Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
+ upon doing any I/O following a direct formatted write.
+
+ (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
+ @code{g77}'s version of @code{libf2c}.)
+ @end itemize
@heading In @code{egcs} 1.1 (versus 0.5.24):
*** g77-e/libf2c/libI77/sfe.c.~1~ Tue May 19 06:51:14 1998
--- g77-e/libf2c/libI77/sfe.c Sat Sep 5 23:03:17 1998
*************** integer e_wsfe(Void)
*** 36,39 ****
--- 36,40 ----
integer e_wdfe(Void)
{
+ f__init = 1;
return en_fio();
}