nslay

Indeed, it's only exact on paper. But it's exact on paper for finitely represented binary numbers using a finite number of operations. Contrast this with infinite series that need infinite...

I think have this registry item turned off already (I've followed that guide before IIRC). I'll have to check. No matter what I do to get rid of CompatTelRunner, it...

From randomly perusing github, I happened on this repository and issue list. I had and solved this problem when I wrapped openslide in ITK. See the comments explaining this and...

This issue is still haunting me in SimpleITK 2.2.0. It doesn't matter if I use sitk.ReadImage, sitk.ImageFileReader or if I manually select an ImageIO. I've also tried newer versions of...

This is a small piece of an image batcher used for training neural networks. There's more following it that post-processes the loaded images, but that's unimportant here. Sorry for all...

Sure. But I want you to understand that this problem is not strictly reproducible. For example, I have hundreds of experiments using 4 image channels (1 nifti file per channel)...

Try the attached out. [SimpleITKCrash.zip](https://github.com/SimpleITK/SimpleITK/files/9576382/SimpleITKCrash.zip) A screenshot is included to show the random nature of the reproducibility. I even added a "No crash!" line later to show successful terminations.

If I change sitk.ReadImage to nib.load, I can't seem to make it crash anymore. Indeed, I have jobs that have been running for a day now without crashing using the...

I don't think that's the problem because memory is only shared within load_one_patient(). Everything returned by load_one_patient is a deep copy. The concatenate is obvious, but the LoadMask is also...

If I had to take a guess... I'd guess this is partly related to factories and their initialization. And ITK loves using lots and lots of factories everywhere!