Thanks for the reply. I've taken the latest release of 1.8.15, and it seems to work, apart from h5dump not finding several files (running make check -i install sees the rest of the checks working fine). I had been using 1.8.9 as I using HDF5 to build NetCDF, and that is the version which was on Unidata's FTP site up, to which one of their build pages still points. I've installed the latest official release, and its behaviour is similar. I'm not sure whether to ignore the missing files, but everything else seems to pass the tests normally.
···
On 22/04/15 20:33, Jerome Soumagne wrote:
Gavin,
I was able to reproduce the segfault but with HDF5 1.8.9, which is the version that it seems you're using (if I look at the trace that you sent), do you have any reasons for sticking with that version? This version is almost 3 years old now.
I gave it another try with the 1.8.15 branch and I can confirm that the test now passes, would you be able to give a try to the 1.8.15 release candidate that has been recently posted?
Thanks
Jerome
On Tue, Apr 21, 2015 at 12:18 PM, Gavin Parnaby <gavin.parnaby@fluidgravity.co.uk><mailto:gavin.parnaby@fluidgravity.co.uk> wrote:
Dear Jerome,
Sorry for the late reply - only just been able to get back onto this. The platform I'm on at the moment is OpenSuSE 13.1 (also x86-64). The kernel version is
3.11.10-25. I'm using the standard configure script, with gcc version 4.8.1, dated 20130909, installed. I'm not passing any compiler options that aren't in the standard script.
The command sequence I'm using to build the libraries is:
./configure --with-zlib=/home/gavin/local --prefix=/home/gavin/local
make check install
I tried running dt_arith on its own, and it indeed segfaulted again. I attach the trace, obtained from gdb.
Thanks very much,
Gavin
On 16/02/15 22:18, Jerome Soumagne wrote:
Hi Gavin,
I just gave it a try on my machine with OpenSuSE 13.2 (x86_64), I cannot
reproduce it. What is your platform? What is the configure line that
you're passing? Are you passing any special compiler options?
Could you please try to run the dt-arith test by itself? from your build
directory, it should be ./test/dt_arith
If it segfaults, could you please run it again with gdb and send a
trace.
Thanks
Jerome
On Fri, 2015-02-06 at 16:52 +0000, Gavin Parnaby wrote:
Hello,
I'm a complete HDF5 newbie, attempting to compile under OpenSuSE 13.1.
I'm able to run the configure script just fine, but the check stage
fails at dt-arith, with the messages:
Testing hard special long double -> float conversions
Child caught signal 11.
Testing hard special long double -> double conversions
Child caught signal 11.
in the test phase. All others were passed.
This led to dt-arithchkexe failing, along with build-check-s, test, and
check-am.
Where might be a good place to look to figure this out?
Thanks very much,
Gavin
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@lists.hdfgroup.org<mailto:Hdf-forum@lists.hdfgroup.org>
http://lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org
Twitter: https://twitter.com/hdf5
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@lists.hdfgroup.org<mailto:Hdf-forum@lists.hdfgroup.org>
http://lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org
Twitter: https://twitter.com/hdf5
