Hi guys! Very interesting idea. I use Python to develop scientific applications. Pythonis great bcoz of the scientific modules but it's a pain to develop applications with it. Particularly, i
I'd love to see:
1) No GIL. There are plenty of solutions for multiprocessing pools for data analytics, but Python is excruciatingly painful when creating concurrent threads that require intercommunication, which is the core of any application. I'd love to be able to create threads easily without having to spawn a whole new process. Also, that it us possible to have memory-shated spaces without the annoyance of using ctypes, which is not native. In reality, I feel thst the true problem here more than the GIL is dynamic typing, but I'm not an expert.
2) Portability of scientific libraries. I know you have it in the pipeline, but thus is the real advantage of python to me.
3) runtime enviornmemt to deploy applications, which is totally lacking.
I know I might be in a very small niche, but I feel there is a need there for building applications that require scientific-level data processing. I know that the standard solution to this problem is:
1) write the app in java/C#/C++ n embed compiled c code from python/Matlab scripts for the data processing bits. But that works well only if the data processing is contained. Otherwise, it's a lot of interface. For a big company, they can afford to have a deducated team for translation, but startups, those that usr python the most, cannot.
2) use scijava or other numerical packages. But those suck.
Python is great because it allows to go from analytics to production, except that production sucks.
Best Regards and Best of luck!
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Hi guys! Very interesting idea. I use Python to develop scientific applications. Pythonis great bcoz of the scientific modules but it's a pain to develop applications with it. Particularly, i
I'd love to see:
1) No GIL. There are plenty of solutions for multiprocessing pools for data analytics, but Python is excruciatingly painful when creating concurrent threads that require intercommunication, which is the core of any application. I'd love to be able to create threads easily without having to spawn a whole new process. Also, that it us possible to have memory-shated spaces without the annoyance of using ctypes, which is not native. In reality, I feel thst the true problem here more than the GIL is dynamic typing, but I'm not an expert.
2) Portability of scientific libraries. I know you have it in the pipeline, but thus is the real advantage of python to me.
3) runtime enviornmemt to deploy applications, which is totally lacking.
I know I might be in a very small niche, but I feel there is a need there for building applications that require scientific-level data processing. I know that the standard solution to this problem is:
1) write the app in java/C#/C++ n embed compiled c code from python/Matlab scripts for the data processing bits. But that works well only if the data processing is contained. Otherwise, it's a lot of interface. For a big company, they can afford to have a deducated team for translation, but startups, those that usr python the most, cannot.
2) use scijava or other numerical packages. But those suck.
Python is great because it allows to go from analytics to production, except that production sucks.
Best Regards and Best of luck!