Re: gcc-2.3.1 -O2 corrupts rc?

Re: gcc-2.3.1 -O2 corrupts rc?

[Byron Rakitzis]

re: gcc. I don't have gcc-2.3 here, you are on your own. Given how
flaky new GNU code is I wouldn't be surprised if it's an optimizer
bug. BTW, you don't mention the architecture you are using.

>P.S. Browsing through the rc-list-file that came with the distribution I have
>     to suppose that the author of rc gave up working on it; archone.tamu.edu
>     seems to have ceased ftp-service.
>     Is there a place where fixes to rc are archived or who (if anybody)
>     maintains rc now?

re: me. I'm still here, as byron@netapp.com. Mail to the old address
*should* be forwarded here. If it isn't, please let me know!

re: fixes. rc-1.5 should be out eventually. Let me get es out of my
hair first. The fact that 1.4 is so stable is a disincentive for me to
work hard on 1.5, btw. However, I am keeping a track of all the (so far
minor) changes that need to be made. Unless I have an unexpected amount
of spare time on my hands, I'm hoping to have something by mid-winter.

re: archives. Chris Siebenmann is of course taking care of this at
Toronto. Thanks Chris! I'm wondering whether a note to this effect
should be posted to the net? i.e., something that will get propagated
to comp.archives.

re: archone.tamu.edu. Apparently Texas A&M University has seen it fit
to ignore all ftp and telnet from off-site. Nice move. I guess it fits
in well with their academic mission, something like that. (Pardon my
cynicism, but this is worse than Stanford turning off fingerd on their
general-access machines (like leyland))

Byron.

gcc-2.3.1 -O2 corrupts rc?

[Stefan Dalibor]

Hello,
interested in finding The Ultimate Shell I felt that I should have a look
at rc; I copied the sources for version 1.4 from archone.tamu.edu, compiled it
with gcc-2.3.1. Compilation made no problems (except warnings:
    proto.h:49: warning: conflicting types for built-in function `memcpy'
    proto.h:49: warning: type mismatch with previous external decl
    <built-in>:0: warning: previous external decl of `memcpy'
which disappeared after removing the forward declaration of memcpy from
proto.h).

But: make trip failed if compiled with option -O2 (all available optimizations
     turned on):
    % make trip
    ./rc -p < trip.rc
    tripping ./rc
      .   .   .   .
(output from successful tests deleted)
      .   .   .   .
    trip took a wrong turn: match
    trip complete
    %

On the other hand, trip.rc completed successfully when rc was compiled with:
    - gcc-2.2.2 -O2
    - gcc-2.3.1 -O
    - cc -Qpath /local/lib/unproto -O[234] -I/local/lib/unproto 
         -Dconst= -Dvolatile=
      where unproto is a package converting ANSI-C to K&R-C by Wietse Venema
      (got it from ftp.win.tue.nl).

Has anybody of you experienced the same? Is this a known bug of gcc-2.3.1 (and
if, has it been reported to GNU)?
Or is it possible that I did something wrong during the installation?

Please e-mail to me, as I'm unsubscribed to this list (I still don't know
whether I'll like rc, who likes shells at all... :-)

P.S. Browsing through the rc-list-file that came with the distribution I have
     to suppose that the author of rc gave up working on it; archone.tamu.edu
     seems to have ceased ftp-service.
     Is there a place where fixes to rc are archived or who (if anybody)
     maintains rc now?

Thanks in advance,
Stefan
----------------------------------------------------
Stefan Dalibor
Lehrstuhl fuer Rechnerstrukturen (IMMD III)
Friedrich-Alexander-Universitaet Erlangen-Nuernberg 
Erlangen, Germany          

e-mail: dalibor@immd3.informatik.uni-erlangen.de