octave-maintainers
[Top][All Lists]
Advanced

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

Re: [RFU] octave-ltfat 2.1.1+dfsg-1


From: Rafael Laboissiere
Subject: Re: [RFU] octave-ltfat 2.1.1+dfsg-1
Date: Mon, 9 Nov 2015 10:02:19 +0100
User-agent: Mutt/1.5.24 (2015-08-30)

[Cc'ing to address@hidden I hope that the Octave developers may have suggestions to fix the problems.]

* Rafael Laboissiere <address@hidden> [2015-11-07 18:36]:

I prepared version 2.1.1+dfsg-1 of the octave-ltfat package in Git (commit f006100). This is an upstream version bump. I removed the architecture restrictions for the unit testing. Please, upload it to unstable and we will on which architectures there will be build failures. Unfortunately, the unit tests in octave-ltfat take a very long time to complete and this may be the cause of the build failures on some old architectures.

Well, according to the buildd web page [1], the octave-ltfat package built correctly on most of the official architectures in Debian unstable. The unit testing failed on ppc64 (powerpc) [2] and s390x [3], and that aborted the build process. The build on mips is takng a very long time to finish and I suspect that it will also fail. At any rate, here are the known reasons of failure:


* ppc64

===============  TEST_PFILT ==============
*** Error in `/usr/bin/octave-cli': munmap_chunk(): invalid pointer: 0x13871e18 
***

This is a memory management problem. This does not happen on other architectures but I am not sure it is related to the s390x architecture or it comes from the specific host used to build the package.


* s390x

===============  TEST_DGT ================
** On entry to ZGEMM  parameter number  8 had an illegal value

This error is related to the call to the zgemm routine of the BLAS library. Since the test passed on all other architectures, I would guess that this is a compiler problem on s390x, but I really do not know.


In the previous releases of the Debian package octave-ltfat, we disabled the unit testing for the problematic architectures, but this is not a good option, because we ship then packages that are potentially buggy on some architectures.

Any help on fixing these problems will be welcome.

Best,

Rafael

 1. https://buildd.debian.org/status/package.php?p=octave-ltfat&suite=unstable
 2. 
https://buildd.debian.org/status/fetch.php?pkg=octave-ltfat&arch=ppc64&ver=2.1.1%2Bdfsg-1&stamp=1446936606
 3. 
https://buildd.debian.org/status/fetch.php?pkg=octave-ltfat&arch=s390x&ver=2.1.1%2Bdfsg-1&stamp=1446934532



reply via email to

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