-
-
Notifications
You must be signed in to change notification settings - Fork 42
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
test_evaluation.py::test_system_specific_long_integer fails on s390x #760
Comments
As you mention, I suspect it is a bigendian difference. What is the output when you run using
? I suspect his will come out |
Here you go:
|
Ok - thanks. Now try changing the failing test to the code in https://github.com/Mathics3/mathics-core/blob/06ae8117f7fcfabcc34855b9db3226b011326f36/test/test_evaluation.py and try running it using using:
If it succeeds, let me know. If not, send the output. It will look something like:
but the dates and numbers will be different. Thanks. |
I kicked off a new test build with this applied and it failed with:
|
Ok - that was a typo on my part (compare with the little-endian output). This now fixed in 27b300a so change the "132" to "134", or pull from that commit, rerun the test and I think this test should now pass. |
Still failing:
|
At least we progressed a couple of tests and are down to the last two (assuming this is the only test that changes due to BigEndian). Commit b56150b now adjusts this and the next test. To speed things up if a test fails just change the number as given by expected and send back the diff. |
Thanks! There was one more missing, commented with the diff on #762. Once that's updated I'll retest again with the new patch to be sure, but we should be getting there. |
Fixed in da1e308 |
Description
When building the latest stable release (5.0.2) on Fedora on s390x (which is a big-endian architecture),
test_evaluation.py::test_system_specific_long_integer
fails withAssertionError: UnsignedInteger32
.How to Reproduce
This unfortunately isn't terribly easy to repro without access to a s390x box (or emulating one with qemu or Hercules, which can be quite involved). Happy to help test a potential fix on your behalf though if needed.
Output Given
Expected behavior
The test should not fail
Your Environment
I can repro this reliably on Fedora Rawhile when building for the s390x architecture. Builds on other architectures all work fine.
Workarounds
I can gate out the test, but that just papers over the problem.
Priority
It's blocking the inclusion of Mathics in Fedora.
Additional context
This came up during the Fedora package review for Mathics.
The text was updated successfully, but these errors were encountered: