Mon 21 Mar 2016 04:40:32 PM UTC, comment #16:
Thanks for your persistence and following up on this. Closing report as invalid as it's a bug in the Intel compiler version 2016.1.150 as confirmed with Intel.
|
Mon 21 Mar 2016 04:13:36 PM UTC, comment #15:
It has been confirmed that this issue is a compiler bug. There is no workaround, and it will be corrected in the next release of the compiler. https://software.intel.com/en-us/comment/1864855#comment-1864855
|
Fri 18 Mar 2016 09:13:55 AM UTC, comment #14:
Unfortunately, the same error. (well, technically the same error):
|
Wed 16 Mar 2016 09:09:26 PM UTC, comment #13:
Ok, I read through the thread and I see some changes were committed to gnulib. I'm attaching an update of the simple test project with those gnulib changes applied and with the GNULIB_NAMESPACE added so you don't need to put it on the command-line now. I'm even adding gnulib's std-gnu11 module (which seems like it is similar to what we're doing in Octave's development branch).
Any different results or still the same error?
(file #36662)
|
Wed 16 Mar 2016 07:09:36 AM UTC, comment #12:
I think it was clear. Anyway, the GNULIB buglist is
http://lists.gnu.org/archive/html/bug-gnulib/2016-03/msg00026.html
Additionally, I also requested Intel C++ support, although not so much activity. (other than acknowledging the problem:
https://software.intel.com/en-us/forums/intel-c-compiler/topic/612811
|
Wed 16 Mar 2016 01:03:00 AM UTC, comment #11:
We are not using gnulib's std-gnu11, I didn't know there was one, we can take a look at that.
For the version of Octave you are building, no compiler flags are added to set the language variant. Also the error you are seeing is when gnulib's sys/select.h is included in C++ code, was that made clear to the gnulib devs? Can you provide a pointer to the gnulib mailing list thread?
|
Tue 15 Mar 2016 09:00:35 PM UTC, comment #10:
I got a question from GNULIB buglist:
Is Octave using Gnulib's std-gnu11 module?
|
Sun 13 Mar 2016 10:22:37 PM UTC, comment #9:
Thanks. Marking this bug as postponed until you can get some resolution from the gnulib team.
|
Sat 12 Mar 2016 06:47:03 AM UTC, comment #8:
Well, we just have confirmed to be as a GNULIB problem. The same error that I got with Octave. (It might also be an Intel compiler problem). I already opened a bug-report with GNULIB, when I will get some info I will share it here. I will also check it with the newest version of Intel compiler.
|
Fri 11 Mar 2016 07:18:25 PM UTC, comment #7:
Sorry, I forgot something that I think is important in this example, please try again with
Does that result in the same error you saw with Octave?
|
Thu 10 Mar 2016 04:46:32 AM UTC, comment #6:
I got a testprog, no errors at all.
|
Wed 09 Mar 2016 10:37:35 PM UTC, comment #5:
To confirm that it's a problem with gnulib, can you try the attached tiny project, do you get the same error?
(file #36592)
|
Wed 09 Mar 2016 09:12:03 PM UTC, comment #4:
Just in case more info:
config.log https://gist.github.com/hajgato/10389cf249832aabb777
config.status https://gist.github.com/hajgato/7f9ff6fa1d1d12bccfe0
config/build output https://gist.github.com/hajgato/bea2eebd8908a86cace2
|
Wed 09 Mar 2016 08:47:14 PM UTC, comment #3:
What I was able to figure out that Intel 2015.1.133 is not affected
|
Wed 09 Mar 2016 08:41:15 PM UTC, comment #2:
I do think it is a gnulib related problem, but I am not mch of a programmer. So I do not know how to proceed. Anyway, I will submit this to the gnulib bugtrack as well.
|
Wed 09 Mar 2016 08:23:53 PM UTC, comment #1:
Thanks for the detailed bug report. If you are interested in helping Octave to be compatible with the Intel compiler, can you try to see if this is a gnulib bug? Can you create a simple test project with gnulib to see if this problem is reproducible, and if so report it to the gnulib project?
|
Wed 09 Mar 2016 08:16:35 PM UTC, original submission:
The error message is:
make[3]: Entering directory `/tmp/hajgato/build/Octave/4.0.0/intel-2016a/octave-4.0.0/liboctave'
CXX array/array_libarray_la-Array-b.lo
icpc: command line warning #10006: ignoring unknown option '-Wold-style-cast'
icpc: command line warning #10006: ignoring unknown option '-Wcast-align'
In file included from /usr/include/sys/types.h(220),
from ../libgnu/sys/types.h(28),
from /usr/include/stdlib.h(320),
from ../libgnu/stdlib.h(36),
from /vsc/apps/brussel/ivybridge/software/GCCcore/4.9.3/bin/../include/c++/4.9.3/cstdlib(72),
from /vsc/apps/brussel/ivybridge/software/GCCcore/4.9.3/bin/../include/c++/4.9.3/bits/stl_algo.h(59),
from /vsc/apps/brussel/ivybridge/software/GCCcore/4.9.3/bin/../include/c++/4.9.3/algorithm(62),
from array/Array.h(32),
from array/Array-b.cc(29):
../libgnu/sys/select.h(598): error: declaration is incompatible with previous "select" (declared at line 109 of "/usr/include/sys/select.h")
_GL_CXXALIAS_SYS (select, int,
^
compilation aborted for array/Array-b.cc (code 2)
make[3]: *** [array/array_libarray_la-Array-b.lo] Error 1
make[3]: Leaving directory `/tmp/hajgato/build/Octave/4.0.0/intel-2016a/octave-4.0.0/liboctave'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/tmp/hajgato/build/Octave/4.0.0/intel-2016a/octave-4.0.0/liboctave'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/tmp/hajgato/build/Octave/4.0.0/intel-2016a/octave-4.0.0'
make: *** [all] Error 2
This is rather similar to this issue (although not the same):
https://savannah.gnu.org/bugs/index.php?43171
|