Hello everyone,
I've recently opened
this issue on Github to alert the original SCOOP author, Yannick Hold, to the fact that the domain registration for
pyscoop.org has expired. I received the following response from Yannick:
---
You can go ahead and register the domain if you'd like.
I believe the parallel programming ecosystem changed tremendously since the inception of SCOOP. When the project was started, the research community faced barren lands in terms of high-level parallel programming that supported common HPC environments. I believe SCOOP did a step in the right direction and lead the way in some aspects. However, this barren land is is no longer reality, as projects like celery, joblib, dask, and even vanilla multiprocessing or pytorch's version are well equipped to deal out-of-the-box or with minimum encumberance with common HPC environments. Furthermore, my research interests changed over the past few years and I doubt I will contribute much to SCOOP anymore.
It was a pleasure to help people and build something useful to me, my colleagues and the research community in general.
I'm hereby starting this thread to continue the discussion over here as prompted by Yannick.
I personally believe that SCOOP is still very much relevant, and I would hate to see it disappear due to lack of maintenance. None of the other libraries that Yannick mentioned as examples provide the ease of use or exact functionality that SCOOP offers with its simple and powerful map implementation. My main use for SCOOP is through our university's HPC cluster running SLURM, and it has increased my code's throughput significantly since I've started using it. However, as the github issues and the messages here show, there are a few bugs that could use some attention.
What I would propose is to organize a volunteer effort to continue working on this project in a fork, and put down a development roadmap for more releases. We can either use the same name or a derivative one, depending on Yannick's preference. There have already been a few posts on Github issues from individuals who have taken the time to fix issues and improve functionality on their own forks, but I think we can achieve greater benefit if we rally together and focus our efforts towards the continued survival of this project. And, I would happily volunteer my time and effort to organize and implement this process.
That said, I would really like to know what the rest of the community thinks.
- What are you currently using SCOOP for? And how easily do you think you can replace it with something else?
- Would you like to see a new SCOOP release soon with updates and bug fixes?
- Are you interested in volunteering as a maintainer? Will you be submitting pull-requests to help out?
Thanks in advance for your responses.
Also, I would like to reiterate my gratitude to Yannick Hold for creating this very useful library. It has definitely saved me personally a lot of time and made my research progress go much faster than it would have without it. Thank you Yannick.
