Hi all,
For several years now I have keeping pari up to date in the
sage-on-gentoo overlay, even getting upstream snapshots as
needed for sage. For a while my stable ebuilds have been picked
up for the main tree by a kind soul.
For several years now I haven't been able to successfully run
any of pari testsuite from portage. And it seems to be something
in the portage environment. I have now decided to seek some help
on the issue as I have run out of things I can think of a long
time ago.
So if I run "ebuild pari-2.11.2.ebuild test" (valid with all older
ebuilds available) I get
>>> Source compiled.
>>> Test phase: sci-mathematics/pari-2.11.2
make -j8 dobench
Making dobench in Olinux-x86_64
make[1]: Entering directory
'/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2/Olinux-x86_64'
* Testing objets gp-dyn..TIME= 1
* Testing analyz gp-dyn..TIME= 3
* Testing number gp-dyn..BUG [10]
* Testing polyser gp-dyn..BUG [3]
* Testing linear gp-dyn..BUG [0]
* Testing elliptic gp-dyn..BUG [14]
* Testing sumiter gp-dyn..TIME= 5
* Testing graph gp-dyn..TIME= 3
* Testing program gp-dyn..TIME= 0
* Testing trans gp-dyn..TIME= 11
* Testing nfields gp-dyn..BUG [38]
+++ [BUG] Total bench for gp-dyn is 57
PROBLEMS WERE NOTED. The following files list them in diff format:
Directory:
/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2/Olinux-x86_64
number-dyn.dif
polyser-dyn.dif
linear-dyn.dif
elliptic-dyn.dif
nfields-dyn.dif
make[1]: *** [Makefile:344: dobench] Error 1
make[1]: Leaving directory
'/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2/Olinux-x86_64'
make: *** [Makefile:34: dobench] Error 2
* ERROR: sci-mathematics/pari-2.11.2::sage-on-gentoo failed (test phase):
* emake failed
*
* If you need support, post the output of `emerge --info
'=sci-mathematics/pari-2.11.2::sage-on-gentoo'`,
* the complete build log and the output of `emerge -pqv
'=sci-mathematics/pari-2.11.2::sage-on-gentoo'`.
* The complete build log is located at
'/var/log/portage/build/sci-mathematics/pari-2.11.2:20190718-215929.log'.
* For convenience, a symlink to the build log is located at
'/dev/shm/portage/sci-mathematics/pari-2.11.2/temp/build.log'.
* The ebuild environment file is located at
'/dev/shm/portage/sci-mathematics/pari-2.11.2/temp/environment'.
* Working directory:
'/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2'
* S: '/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2'
fbissey@moonloop ~/Work/Overlays/sage-on-gentoo/sci-mathematics/pari $
cd /dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2
fbissey@moonloop
/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2 $ make -j8
dobench
Making dobench in Olinux-x86_64
make[1]: Entering directory
'/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2/Olinux-x86_64'
* Testing objets gp-dyn..TIME= 0
* Testing analyz gp-dyn..TIME= 0
* Testing number gp-dyn..TIME= 1
* Testing polyser gp-dyn..TIME= 3
* Testing linear gp-dyn..TIME= 0
* Testing elliptic gp-dyn..TIME= 14
* Testing sumiter gp-dyn..TIME= 5
* Testing graph gp-dyn..TIME= 3
* Testing program gp-dyn..TIME= 3
* Testing trans gp-dyn..TIME= 11
* Testing nfields gp-dyn..TIME= 42
+++ Total bench for gp-dyn is 48
make[1]: Leaving directory
'/dev/shm/portage/sci-mathematics/pari-2.11.2/work/pari-2.11.2/Olinux-x86_64'
That's right the tests fails in the ebuild but pass when I run the same
command manually/ It is not a fluke due to running the tests twice.
If I do "ebuild ... compile" and then run the tests manually, they will
succeed.
When I look at the diff produced by the test suite I gat stuff like
*** ../src/test/32/number 2018-07-26 06:00:21.000000000 +1200
--- gp.out 2019-07-19 10:11:20.505605285 +1200
***************
*** 60,70 ****
? factor(100!+1,0)
! [101 1]
! [14303 1]
! [149239 1]
[432885273849892962613071800918658949059679308685024481795740765527568493010
727023757461397498800981521440877813288657839195622497225621499427628453 1]
--- 60,73 ----
? factor(100!+1,0)
! [
!
101 1]
! [
!
14303 1]
! [
!
149239 1]
[432885273849892962613071800918658949059679308685024481795740765527568493010
727023757461397498800981521440877813288657839195622497225621499427628453 1]
The only difference is a widely different formatting of the output when
run under the portage environment.
Does anyone have a clue as to how this is possible? What could cause it
and more importantly, how to fix it.
Cheers,
François
