-
-
Notifications
You must be signed in to change notification settings - Fork 23
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix build when SIGSTKSZ is not a constant. #151
Conversation
…ise in case of OSX bug
I used kliem:fix_non_constant_MINSIGSTKSZ to creat a cysignals package to be able to compile Sagemath in Fedora 35. The test of |
I requested the fixes to cysignals-CSI already in #149. |
A test of a giac file did not pass, with a complain to cysignals but I do not know if it is related with this fix. ./sage -t --long --warn-long 131.6 --random-seed=0 src/sage/libs/giac/init.py File "src/sage/libs/giac/init.py", line 18, in sage.libs.giac |
I think the giac failure is not related. It's just wrapped in |
However, it also makes sense to have the fixes here. #149 is not as urgent as this issue. |
should we merge this? |
I think so and make a release pretty soon. I just closed #139, so that we can run some tests. |
The testsuite fails apparently. Have to wait until the artifacts are uploaded. More importantly my test that I added to configure does not work at all. Any help is welcome. |
What's the branch you're testing? |
Some of the runs seem to work, no? |
Yes, but |
I can confirm that it is possible to build a running copy of sage (from 9.4 with python 3.9 and from develop with python 3.10). There are more tickets involved (to be able to build in F35) and some tests fail, but I can use it in the daily work. |
I believe that the current state of the branch is not working, because You can try this with cloning this rep and just run
Not what I was aiming for, because on Fedora 35 it is not constant. On the other hand, if I do slight modifications, it fails for me, e.g.
So I'm gathering that it is not testing, what it is supposed to test. |
I just bundled your branch and pass it to the build of sage in F35 (I got errors if I build it directly, there are some hacks in spkg-install.in |
I do not understand what happens, I have errors when building outside sage; I attach the log of the sage installation |
Does the latest push work for you. As you probably guessed, your config.log should say: |
Have you checked the patch in the src.rpm package for Fedora 35. There are several ones, maybe this one is relevant. It builds, though using your configure.ac, it still says it is a constant. --- src/cysignals/implementation.c.orig 2021-05-27 14:03:35.653774857 -0600
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
it seems some pairs of [ ]
are missing in the test program. Moreover iirc include statements come apart from the rest of the source, like in a different parameter?
Thank you. This seems to be exactly the problem: Edit: Should have checked with
Of course this is going to work independent of whether MINSIGSTKSZ is a constant. |
This reverts commit 1187e24.
Now it builds for me in Fedora 35. I will try now inside sage. Thanks! |
As can be observed by the checks of #152, the changes here did not break anything. But we do fix Fedora 35 and other new systems. |
A new release then? |
Yes, I'm just trying to figure out a two bug fixes yet. The macos test suite doesn't work at all and ubuntu-trusty fails. If there are no easy solutions to them, I'll just leave it at that. I'm running the doctests verbose to see if I can figure out the ubuntu-trusty issue: https://github.com/kliem/cysignals/actions/runs/1498523090 The original logs were not useful at all:
#154 attempts to fix the mac test suite problem. If it doesn't work, it doesn't work. |
Does this happen locally on macOS too, or it's GitHub actions-only problem?
…On Wed, 24 Nov 2021, 09:17 Jonathan Kliem, ***@***.***> wrote:
Yes, I'm just trying to figure out a two bug fixes yet. The macos test
suite doesn't work at all and ubuntu-trusty fails. If there are no easy
solutions to them, I'll just leave it at that.
I'm running the doctests verbose to see if I can figure out the
ubuntu-trusty issue:
https://github.com/kliem/cysignals/actions/runs/1498523090
The original logs were not useful at all:
ulimit 2>/dev/null -s 1024; python3 -B rundoctests.py src/cysignals/*.pyx
cd example && python3 setup.py clean build
Doctesting 5 files.
src/cysignals/alarm.pyx
running clean
running build
Compiling cysignals_example.pyx because it changed.
[1/1] Cythonizing cysignals_example.pyx
running build_ext
building 'cysignals_example' extension
creating build
creating build/temp.linux-x86_64-3.9
gcc -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wno-unused -g -O2 -g -O2 -Wp,-U_FORTIFY_SOURCE -fPIC -I/sage/local/var/lib/sage/venv-python3.9.7/lib/python+++3.9/site-packages/cysignals -I/sage/local/var/lib/sage/venv-python3.9.7/include/python3.9 -c cysignals_example.cpp -o build/temp.linux-x86_64-3.9/cysignals_example.o
creating build/lib.linux-x86_64-3.9
g++ -shared -Wl,-rpath-link,/sage/local/lib -L/sage/local/lib -Wl,-rpath,/sage/local/lib -Wl,-rpath-link,/sage/local/lib -L/sage/local/lib -Wl,-rpath,/sage/local/lib -Wl,-rpath+++-link,/sage/local/var/lib/sage/venv-python3.9.7/lib -Wl,-rpath,/sage/local/var/lib/sage/venv-python3.9.7/lib -L. -L/sage/local/lib -Wl,-rpath-link,/sage/local/lib -L/sage/local/+++lib -Wl,-rpath,/sage/local/lib -Wl,-rpath-link,/sage/local/lib -L/sage/local/lib -Wl,-rpath,/sage/local/lib -Wl,-rpath-link,/sage/local/lib -L/sage/local/lib -Wl,-rpath,/sage/lo+++cal/lib -Wl,-rpath-link,/sage/local/lib -L/sage/local/lib -Wl,-rpath,/sage/local/lib -g -O2 -Wp,-U_FORTIFY_SOURCE build/temp.linux-x86_64-3.9/cysignals_example.o -L/sage/local/v+++ar/lib/sage/venv-python3.9.7/lib -o build/lib.linux-x86_64-3.9/cysignals_example.cpython-39-x86_64-linux-gnu.so -lpari
src/cysignals/pselect.pyx
src/cysignals/pysignals.pyx
src/cysignals/signals.pyx
src/cysignals/tests.pyx
make[3]: *** [check-doctest] Error 1
make[3]: Target `check-install' not remade because of errors.
#154 <#154> attempts to fix the
mac test suite problem. If it doesn't work, it doesn't work.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#151 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJXYHFI3GTTXQLXJE2OO33UNSURXANCNFSM5E47TOBQ>
.
|
I don't know. I don't have a macOS. |
This is intended to fix #150 as done in rr-debugger/rr#2916.