Re: Typename5.C failure i386-linux-gnulibc1 [Re: egcs-971127: testsuiteresults on i586-pc-linux-gnulibc1]

Re: Typename5.C failure i386-linux-gnulibc1 [Re: egcs-971127: testsuiteresults on i586-pc-linux-gnulibc1]

[Max Lawson]

	Maybe, we should post assembly and/or preprocessed outputs for 
a given test-case compiled with given options. This way s.o. :-) could 
have a look... Could s.o. post an adequate test ?

Regards, Max
 

Re: egcs-971127: testsuite results on i586-pc-linux-gnulibc1

[Theodore Papadopoulo]

> 
> Max Lawson <mlawson@drfmc.ceng.cea.fr> writes:
> > 
> > 	=> typename5.C is still failing. That was the case in the 
> > previous snapshot. I've added the log. Case ...
> > 
> 
> I don't see the error on my system (RH 4.1, libc 5.3.12), and I didn't
> see it for the last snapshot either.
> 

	Mummit,

From the other reports I see, you seem more to be the exception than
the average case or maybe your are the only one with RH4.1 which seems
unlikely.

I also have:

XPASS: g++.jason/destruct3.C - (test for bogus messages, line 38)
XPASS: g++.mike/dyncast1.C  Execution test
XPASS: g++.mike/dyncast2.C  Execution test
FAIL: g++.pt/typename5.C (test for excess errors)
 
=== g++ Summary ===

# of expected passes            3399
# of unexpected failures        1
# of unexpected successes       3
# of expected failures          80
# of untested testcases         6
/0/robotvis/papadop/gcc/egcs/egcs-971127/gcc/testsuite/../xgcc version egcs-2.90.19 971127 (gcc2-970802 experimental)


My system is Slackware based with a kernel Linux 2.0.29. with two libs
/lib/libc.so.4.7.6 and /lib/libc.so.5.3.12. As for binutils, I'm
using binutils-2.8.1.0.15.tar.gz but I cannot install them so that I'm
just putting the directory binutils into the egcs-971127. I assume this
works as exception handling works fine !!

	Theo.


--------------------------------------------------------------------
Theodore Papadopoulo
Email:  papadop@sophia.inria.fr               Tel: (33) 93 65 76 01
--------------------------------------------------------------------

Typename5.C failure i386-linux-gnulibc1 [Re: egcs-971127: testsuiteresults on i586-pc-linux-gnulibc1]

[Mumit Khan]

On Mon, 1 Dec 1997, Theodore Papadopoulo wrote:

> From the other reports I see, you seem more to be the exception than
> the average case or maybe your are the only one with RH4.1 which seems
> unlikely.
> 

Nope, I'm not the only one. See 

    http://www.cygnus.com/ml/egcs/1997-Nov/0946.html          

for the results from Peter Schmid, posted on Nov. 29. He also does not see
the failure for typename5.C that you see. Max Lawson's message that shows 
the error message from running the compiler on typename5.C just doesn't
happen on my machine.

Mumit

Re: Typename5.C failure i386-linux-gnulibc1 [Re: egcs-971127: testsuiteresults on i586-pc-linux-gnulibc1]

[Robert Lipe]

> > From the other reports I see, you seem more to be the exception than
> > the average case or maybe your are the only one with RH4.1 which seems
> 
> for the results from Peter Schmid, posted on Nov. 29. He also does not see
> the failure for typename5.C that you see. Max Lawson's message that shows 


Perhaps there's some sort of slippery x86 thing going on here, becuase
typename5.C fails on i586-pc-sco3.2v5.0.4, too.   See:
	http://www.cygnus.com/ml/egcs/1997-Nov/0950.html


Interestingly, it seems to be a recent failure.

$ grep typename5.C */*.sum
coff-971123/g++.sum:PASS: g++.pt/typename5.C (test for excess errors)
coff-971129/g++.sum:FAIL: g++.pt/typename5.C (test for excess errors)
elf-971123/g++.sum:PASS: g++.pt/typename5.C (test for excess errors)
elf-971129/g++.sum:FAIL: g++.pt/typename5.C (test for excess errors)

Re: Typename5.C failure i386-linux-gnulibc1 [Re: egcs-971127: testsuiteresults on i586-pc-linux-gnulibc1]

[Mumit Khan]

Robert Lipe <robertl@dgii.com> writes:
> 
> Perhaps there's some sort of slippery x86 thing going on here, becuase
> typename5.C fails on i586-pc-sco3.2v5.0.4, too.   See:
> 	http://www.cygnus.com/ml/egcs/1997-Nov/0950.html
> 
> 
> Interestingly, it seems to be a recent failure.
> 
> $ grep typename5.C */*.sum
> coff-971123/g++.sum:PASS: g++.pt/typename5.C (test for excess errors)
> coff-971129/g++.sum:FAIL: g++.pt/typename5.C (test for excess errors)
> elf-971123/g++.sum:PASS: g++.pt/typename5.C (test for excess errors)
> elf-971129/g++.sum:FAIL: g++.pt/typename5.C (test for excess errors)
> 

To add another data point, see Alexandre Oliva's test results just posted.
He's getting lots of EH failures which points to the fact that he may be
using an old gas, yet *no* typename5.C failure.

This is certainly getting weird. fyi, I couldn't reproduce the problem on
any ix86 system I have access to -- linux-gnulibc1, cygwin32 and mingw32,
all running the same gas/binutils snapshot (binutils ss-971023).

Regards,
Mumit -- khan@xraylith.wisc.edu
http://www.xraylith.wisc.edu/~khan/