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
Regression blocking Debian #1264
Comments
@rpavlik So is this is a custom Debian test suite/script using the files from our test suite and solvespace-cli ? I am not on Linux but I did run the equivalent of the last command
On Windows:
and it exported a valid step file. This was in debug mode. Here is the output:
The same in release:
And the output:
|
yeah, it only shows up on the s390x architecture which is IBM z/Series aka system 360 successor. I have a docker environment setting up on my pc (using binfmt and qemu) now to bisect here's the script: https://salsa.debian.org/science-team/solvespace/-/blob/master/debian/tests/surfaces (also those are very weird mimalloc warnings...) |
OK, so I'm not seeing the error in a release build of master inside an s390x docker... |
Call me very impressed that you know how to set up and IBM s/390 emulation with qemu and docker! Crazy thought - if you run solvespace (the GUI) and point it to an X server on a PC would it actually work? The s/390 does not have "local" display hardware I presume? |
Yeah , x forwarding would probably work. I had never done it either, and am frankly surprised it works too. My first try was to install a vm, but I found instructions to do it with docker and it worked, so... I am now basically trying to rerun the tests: on testing (bookworm), with the unstable (sid) version of the solvespace package installed (3.1 basically), doing the autopkgtests. Some weird results, like:
which nevertheless actually did not fail the test? I replied to the bug requesting a re-test. |
btw # does some kind of setup?
docker run --rm --privileged multiarch/qemu-user-static --reset -p yes
docker run --rm -it --platform linux/s390x s390x/debian:bookworm |
I ran master on x86-64 linux, opened the test file with the GUI and exported surfaces. Got what looks like a valid .step file. |
Yeah, see the linked Debian bug: it's a very specific environment causing failure (architecture, system deps, and maybe cosmic rays). It works fine on more common setups and on my attempt to replicate the failed setup. |
@rpavlik now there is a new (semi?)automatic bug: |
Yeah I have to email the s390 mailing list. If they can't help I'll probably just remove the package from that architecture, it's pretty clearly a spurious problem, and I'm not sure if anybody is actually running Linux on Z series mainframes to use GUI apps like Solvespace 🤣 |
I wanted to suggest that, but who am I to say what architectures Debian folks want solvespace to run on or why? It is the obvious solution, and I doubt anyone is using it for actual CAD work on that arch. |
@rpavlik ?? |
Yep, I got that email, plan to address it this week at work. |
OK, I am now just waiting for a sponsorship of the revised package, since I am not formally a DM or DD. The underlying float precision or something issue on s390x remains, but at least this should unblock debian. |
@rpavlik It seems 3.1 is in Debian now https://tracker.debian.org/pkg/solvespace - can we close this? |
yeah, it's not blocking debian anymore. It's not a build issue, unless there are flags to make s390x floating point act more like sse2. |
System information
Expected behavior
autopkgtests run normally
Actual behavior
"no surfaces to export" on s390x: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1013163
and
This did work with the previous Debian package version, which is a 3.0RC.
The text was updated successfully, but these errors were encountered: