GNU bug report logs - #65437
bloomberg-bde 3.98.0 non-determistic test failure

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 65437 in the body.
You can then email your comments to 65437 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
To: bug-guix <bug-guix <at> gnu.org>
Subject: bloomberg-bde 3.98.0 non-determistic test failure
Date: Mon, 21 Aug 2023 12:19:10 -0400

Hello,

Seen on Cuirass:

--8<---------------cut here---------------start------------->8---
 41%] Built target bslmf_ismemberpointer.t
[ 41%] Linking CXX executable bslmf_ismemberfunctionpointer.t
/gnu/store/ygab8v4ci9iklaykapq52bfsshpvi8pw-cmake-minimal-3.24.2/bin/cmake -E cmake_link_script CMakeFiles/bslmf_ismemberfunctionpointer.t.dir/link.txt --verbose=1
/gnu/store/5lqhcv91ijy82p92ac6g5xw48l0lwwz4-gcc-11.3.0/bin/c++ -O2 -g -DNDEBUG CMakeFiles/bslmf_ismemberfunctionpointer.t.dir/groups/bsl/bslmf/bslmf_ismemberfunctionpointer.t.cpp.o -o bslmf_ismemberfunctionpointer.t  -Wl,-rpath,/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build libbslmf.a libbsla.a libbslscm.a libbsls.a -lrt libryu.so 
make[3]: Leaving directory '/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build'
[ 41%] Built target bslmf_ismemberfunctionpointer.t
make[2]: Leaving directory '/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build'
make[1]: *** [CMakeFiles/Makefile2:40289: CMakeFiles/all.t.dir/rule] Error 2
make[1]: Leaving directory '/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build'
make: *** [Makefile:14690: all.t] Error 2
error: in phase 'build-tests': uncaught exception:
%exception #<&invoke-error program: "make" arguments: ("all.t" "-j" "16") exit-status: 2 term-signal: #f stop-signal: #f> 
phase `build-tests' failed after 705.6 seconds
command "make" "all.t" "-j" "16" failed with status 2
builder for `/gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv' failed with exit code 1
@ build-failed /gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv - 1 builder for `/gnu/store
--8<---------------cut here---------------end--------------->8---

The build was broken on the qt-updates branch [0] but working on master, so
it appears to be non-deterministic.

[0]  https://ci.guix.gnu.org/build/991957/log/raw

-- 
Thanks,
Maxim

Message #8 received at 65437 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
Cc: 65437 <at> debbugs.gnu.org, Greg Hogan <code <at> greghogan.com>
Subject: Re: bug#65437: bloomberg-bde 3.98.0 non-determistic test failure
Date: Tue, 22 Aug 2023 23:54:44 +0200

Hi,

Maxim Cournoyer <maxim.cournoyer <at> gmail.com> skribis:

> make: *** [Makefile:14690: all.t] Error 2
> error: in phase 'build-tests': uncaught exception:
> %exception #<&invoke-error program: "make" arguments: ("all.t" "-j" "16") exit-status: 2 term-signal: #f stop-signal: #f> 
> phase `build-tests' failed after 705.6 seconds
> command "make" "all.t" "-j" "16" failed with status 2
> builder for `/gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv' failed with exit code 1
> @ build-failed /gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv - 1 builder for `/gnu/store
>
> The build was broken on the qt-updates branch [0] but working on master, so
> it appears to be non-deterministic.
>
> [0]  https://ci.guix.gnu.org/build/991957/log/raw

The actual error seems to hide above in the log:

--8<---------------cut here---------------start------------->8---
[ 28%] Linking CXX executable bslma_managedptr_pairproxy.t
/gnu/store/ygab8v4ci9iklaykapq52bfsshpvi8pw-cmake-minimal-3.24.2/bin/cmake -E cmake_link_script CMakeFiles/bslma_managedptr_pairproxy.t.dir/link.txt --verbose=1
/gnu/store/5lqhcv91ijy82p92ac6g5xw48l0lwwz4-gcc-11.3.0/bin/c++ -O2 -g -DNDEBUG CMakeFiles/bslma_managedptr_pairproxy.t.dir/groups/bsl/bslma/bslma_managedptr_pairproxy.t.cpp.o -o bslma_managedptr_pairproxy.t  -Wl,-rpath,/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build libbslma.a libbslmf.a libbsla.a libbslscm.a libbsls.a -lrt libryu.so 
/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp: In function ?int main(int, char**)?:
/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp:11176:22: error: ?move? is not a member of ?std?
11176 |             o = std::move(o2);
      |                      ^~~~
/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp:26:1: note: ?std::move? is defined in header ?<utility>?; did you forget to ?#include <utility>??
   25 | #include <stdlib.h>     // 'atoi'
  +++ |+#include <utility>
   26 | #include <string.h>
make[3]: Leaving directory '/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build'
--8<---------------cut here---------------end--------------->8---

I’m guessing that those .t.cpp files are generated and then built, and
that there’s a race condition when building in parallel.

Greg, does that ring a bell?  Should we perform the ‘build-tests’ phase
sequentially?

Ludo’.

Message #11 received at 65437 <at> debbugs.gnu.org (full text, mbox):

From: Greg Hogan <code <at> greghogan.com>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: 65437 <at> debbugs.gnu.org, Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
Subject: Re: bug#65437: bloomberg-bde 3.98.0 non-determistic test failure
Date: Tue, 15 Jul 2025 13:01:34 -0400

On Tue, Aug 22, 2023 at 5:54 PM Ludovic Courtès <ludo <at> gnu.org> wrote:
>
> Hi,
>
> Maxim Cournoyer <maxim.cournoyer <at> gmail.com> skribis:
>
> > make: *** [Makefile:14690: all.t] Error 2
> > error: in phase 'build-tests': uncaught exception:
> > %exception #<&invoke-error program: "make" arguments: ("all.t" "-j" "16") exit-status: 2 term-signal: #f stop-signal: #f>
> > phase `build-tests' failed after 705.6 seconds
> > command "make" "all.t" "-j" "16" failed with status 2
> > builder for `/gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv' failed with exit code 1
> > @ build-failed /gnu/store/sy9glkyrda4na297m9v85i53png5pvbh-bloomberg-bde-3.98.0.0.drv - 1 builder for `/gnu/store
> >
> > The build was broken on the qt-updates branch [0] but working on master, so
> > it appears to be non-deterministic.
> >
> > [0]  https://ci.guix.gnu.org/build/991957/log/raw
>
> The actual error seems to hide above in the log:
>
> --8<---------------cut here---------------start------------->8---
> [ 28%] Linking CXX executable bslma_managedptr_pairproxy.t
> /gnu/store/ygab8v4ci9iklaykapq52bfsshpvi8pw-cmake-minimal-3.24.2/bin/cmake -E cmake_link_script CMakeFiles/bslma_managedptr_pairproxy.t.dir/link.txt --verbose=1
> /gnu/store/5lqhcv91ijy82p92ac6g5xw48l0lwwz4-gcc-11.3.0/bin/c++ -O2 -g -DNDEBUG CMakeFiles/bslma_managedptr_pairproxy.t.dir/groups/bsl/bslma/bslma_managedptr_pairproxy.t.cpp.o -o bslma_managedptr_pairproxy.t  -Wl,-rpath,/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build libbslma.a libbslmf.a libbsla.a libbslscm.a libbsls.a -lrt libryu.so
> /tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp: In function ?int main(int, char**)?:
> /tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp:11176:22: error: ?move? is not a member of ?std?
> 11176 |             o = std::move(o2);
>       |                      ^~~~
> /tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/source/groups/bsl/bslma/bslma_managedptr.t.cpp:26:1: note: ?std::move? is defined in header ?<utility>?; did you forget to ?#include <utility>??
>    25 | #include <stdlib.h>     // 'atoi'
>   +++ |+#include <utility>
>    26 | #include <string.h>
> make[3]: Leaving directory '/tmp/guix-build-bloomberg-bde-3.98.0.0.drv-0/build'
> --8<---------------cut here---------------end--------------->8---
>
> I’m guessing that those .t.cpp files are generated and then built, and
> that there’s a race condition when building in parallel.
>
> Greg, does that ring a bell?  Should we perform the ‘build-tests’ phase
> sequentially?
>
> Ludo’.

Without digging too deeply into old code, the files are not generated
[0] and this error looks to have been a missing header which was long
ago added [1]. Unclear why this issue was non-deterministic.

CI has been nearly non-functional but I have not experienced this
build issue in the ~2 years since the error was reported. The proposed
change is trivial but does greatly slow down a lengthy build.

We have had non-deterministic test failures with this package, so I
have updated it on the c++-team branch along with a patch to enable by
default ctest's option to repeat failed tests [2] (which I set to 5
failures by default).

Can we close this issue and follow up as necessary on codeberg?

Greg

[0] https://github.com/bloomberg/bde/blob/3.98.0.0/groups/bsl/bslma/bslma_managedptr.t.cpp
[1] https://github.com/bloomberg/bde/commit/eaeae744c3f99ba9590f9690565173834d2606bc#diff-5e3d1c0e05fe78822b49b9c81cdba7ad62c92741874188686e0063e8c2987ae6
[2] https://cmake.org/cmake/help/latest/manual/ctest.1.html#cmdoption-ctest-repeat

Message #16 received at 65437-done <at> debbugs.gnu.org (full text, mbox):

From: Andreas Enge <andreas <at> enge.fr>
To: 65437-done <at> debbugs.gnu.org
Subject: Close
Date: Tue, 15 Jul 2025 23:43:52 +0200

I agree we should close this, done!

Thanks for looking into it,

Andreas

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.