Incompatible numpy versions

[NathanTambeurRelu]

Hey,

Thank you for the great library!
I was trying to integrate it into our codebase but noticed that it only works for numpy versions 1.23 or lower.
Using our numpy version 1.26.4 I get the following error:

[2024-08-30 09:49:26.528] [CoACD] [info] iter 5 ---- waiting pool: 32
[2024-08-30 09:49:26.528] [CoACD] [info] Processing [0.0%]
[2024-08-30 09:49:26.528] [CoACD] [info] Processing [75.0%]
[2024-08-30 09:49:26.528] [CoACD] [info] Processing [37.5%]
[2024-08-30 09:49:26.528] [CoACD] [info] Processing [87.5%]
[2024-08-30 09:49:26.729] [CoACD] [info] Processing [12.5%]
[2024-08-30 09:49:26.918] [CoACD] [info] Processing [25.0%]
[2024-08-30 09:49:26.987] [CoACD] [info] Processing [50.0%]
[2024-08-30 09:49:27.249] [CoACD] [info] Processing [62.5%]
python: /project/3rd/cdt/CDT/include/CDTUtils.hpp:188: CDT::Index CDT::edgeNeighborInd(const VerticesArr3&, CDT::VertInd, CDT::VertInd): Assertion `vv[0] == iVedge1 || vv[1] == iVedge1 || vv[2] == iVedge1' failed.

Any advice on how I could fix this?

[SarahWeiii]

This seems to be an error from CDT, not from numpy, does it still happen if you run again on the same input?

[NathanTambeurRelu]

It happens every time I run the algorithm at the exact same moment in the algorithm. The cause is indeed cdt not working with numpy versions above 1.23 I think.