docly icon indicating copy to clipboard operation
docly copied to clipboard

Published 20 hours ago • verified publisher icon autosoft-dev

What does Segmentation fault mean?

Open hyeon95y opened this issue 4 years ago • 4 comments

image

  • It fails on specific file, what does it mean?
  • If some file fails, the whole procedure stop when it's applied on a directory. I think it should run anyway through other files whether if failed on specific file or not

hyeon95y avatar Feb 04 '21 08:02 hyeon95y

Segmentation fault generally comes from libraries and function calls closer to the system. I am afraid that from the Python level we won't have much to do. That said I experienced that yesterday in a brand new Ubuntu 20.04 machine with GPU with python 3.8 and Pytorch 1.6+cpu version. I started to look into the same issue. Can you tell me a bit about your configuration please?

Hope that we will find a solution easily.

(to give a better perspective, in this colab - https://colab.research.google.com/drive/19O7uvH2bZ-K0DGxnSW3_C7B4tzzvWUsT?usp=sharing the segmentation fault does not occur on same file for which I saw it happening in the machine I mentioned earlier.)

rcshubhadeep avatar Feb 04 '21 09:02 rcshubhadeep

Thank you for your answer. I was working in docker image (rayproject/ray-ml:1.0.0), but not exactly same image from the docker hub as I installed more packages.

I also thought about just showing pip list, but it seems not a good way. If there's any formal way to share configuration, please tell me and I'll show you my configuation.

Furthermore, I think some non-english characters in the code may cause the error. As I'm reading the code from the other person for the first time for refactoring, I cannot say the code itself is perfect and may not cause any syntax, or any unknown error. I'll also tell you If I find something while refactoring that.

hyeon95y avatar Feb 04 '21 12:02 hyeon95y

Ok great. Thanks. I hope that docly will be helpful for you. If you need to go ahead then clone the colab I linked in my last message and then you can upload some code files and run docly in the colab environment. That is not the optimal solution but until I find a fix that is a good way to go.

Anyway, it is nice to see that you consider docly in a real use case. Always happy to help

rcshubhadeep avatar Feb 04 '21 13:02 rcshubhadeep

Thanks for your support. I'll share any information as it comes out. Hope your app become new standard like black, isort :)

hyeon95y avatar Feb 05 '21 03:02 hyeon95y