octave-maintainers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Octave 4.0.0 release candidate 1 available for ftp


From: Orion Poplawski
Subject: Re: Octave 4.0.0 release candidate 1 available for ftp
Date: Tue, 10 Mar 2015 16:06:31 -0600
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.5.0

On 03/10/2015 01:57 PM, Dmitri A. Sergatskov wrote:
> 
> 
> On Tue, Mar 10, 2015 at 1:03 PM, Orion Poplawski <address@hidden
> <mailto:address@hidden>> wrote:
> 
> 
>     octave:4> a =  1 + 1i
>     a =  1 + 1i
>     octave:5> b =  1 + 1i
>     b =  1 + 1i
>     octave:6> b < a
>     ans =  1
>     octave:7> b > a
>     ans = 0
>     octave:8> b == a
>     ans =  1
> 
>     Any idea where to look next?
> 
> 
> 
> ​Seeing that this was compiled by gcc 5.0 pre-release, I have a strong
> suspicion this is a gcc/stdlib issue.
> Perhaps you could circumvent it with -mfpmath=sse option to gcc.​
> 
> ​Dmitri.

Not an unrealistic assumption.  Especially since we are now seeing the same
failure building the current octave 3.8.2 rpm with gcc 5.0.0.  Adding
-mfpmath=sse, and changing -march=i686 to -march=atom, so:

  C compiler:                  gcc  -pthread -fopenmp -mieee-fp  -Wall -W
-Wshadow -Wformat -Wpointer-arith -Wmissing-prototypes -Wstrict-prototypes
-Wwrite-strings -Wcast-align -Wcast-qual -O2 -g -pipe -Wall
-Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fexceptions
-fstack-protector-strong --param=ssp-buffer-size=4 -grecord-gcc-switches
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m32 -march=atom -mtune=atom
-fasynchronous-unwind-tables -mfpmath=sse
  C++ compiler:                g++  -pthread -fopenmp -mieee-fp  -Wall -W
-Wshadow -Wold-style-cast -Wformat -Wpointer-arith -Wwrite-strings
-Wcast-align -Wcast-qual -O2 -g -pipe -Wall -Werror=format-security
-Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
--param=ssp-buffer-size=4 -grecord-gcc-switches
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m32 -march=atom -mtune=atom
-fasynchronous-unwind-tables -mfpmath=sse

This does fix the problem.  Unfortunately, this is not acceptable for Fedora
packages as we can't assume sse.

Can someone point me to the code that would be running this comparison so I
can run it through the debugger?

Thanks,

   Orion

-- 
Orion Poplawski
Technical Manager                     303-415-9701 x222
NWRA, Boulder/CoRA Office             FAX: 303-415-9702
3380 Mitchell Lane                       address@hidden
Boulder, CO 80301                   http://www.nwra.com



reply via email to

[Prev in Thread] Current Thread [Next in Thread]