search

ANTsX / ANTsPy

A fast medical imaging analysis library in Python with algorithms for registration, segmentation, and more.
https://antspyx.readthedocs.io
Apache License 2.0
625 stars 161 forks source link

Failing tests #461

Closed cookpa closed 1 year ago

cookpa commented 1 year ago

Hi @ntustison ,

Lots of test failures on the most recent PR.

https://github.com/ANTsX/ANTsPy/actions/runs/4969163765/jobs/8892176590

Errors about transform factories - do you think these could be related to the recent troubles with ANTsR? But I don't get why this is suddenly a thing. The runners should be the same since the switch to Ubuntu 22.04 in October 2022 (https://github.com/actions/runner-images/issues/6399). The ANTs version has not changed since #442. But there are some settings in the runners that might change. I'll see what I can do to help lock those down.

Did your recent PR build successfully on your machine? Would you mind sharing the details of what system, compiler you used?

ntustison commented 1 year ago

No, I'm pretty sure this stems from this commit from @richardbeare which expands template building to include the linear averaging step. The problem is that the original code from Gang uses the MatrixOffsetTransformBase class which, I learned from Hans a long time ago, was never meant to actually be used which is why it's not registered with the other transform factories. So it's unrelated but it still needs to get fixed. I'll try to look at it later.

ntustison commented 1 year ago

I think it's here. Let me see if that fixes it.

ntustison commented 1 year ago

https://github.com/ANTsX/ANTsPy/pull/462

cookpa commented 1 year ago

Thanks - hopefully that does it

ntustison commented 1 year ago

Closing as it looks like the PR fixed the issue. Thanks @cookpa .

cookpa commented 1 year ago

Thanks! FYI Windows tests always pass, they need some additional coding to exit with failure. But everything looks good now.

richardbeare commented 1 year ago

Sorry! :facepalm:

cookpa commented 1 year ago

Sorry! :facepalm:

It's all good - thanks for the new functionality