3 ms·

This set of jupyter + autoreloads + editor features might get you close, but you won't be on par to CL's level. I'm thinking about CL's interactive debugger, an

This set of jupyter + autoreloads + editor features might get you close, but you won't be on par to CL's level. I'm thinking about CL's interactive debugger, and about language features that were thought with interactivity in mind. On an error, you get the interactive debugger, you can explore the stack frames and get to the erroneous function, change it, recompile it, come back to the debugger and ask it to resume progress from this point, and see execution pass. You didn't have to re-run the whole operation. If we can do that in Python, I've been in my cave for too long^^ As such, this is how debugging a complex program looks like in Python vs in CL: https://lisp-journey.gitlab.io/images/debugging-python-VS-lisp.png https://lisp-journey.gitlab.io/images/debugging-python-VS-li... (Ravenpack [1] is a company that does big data analysis and that uses CL, they gave some talks, maybe you'll find more practical reasons from them) for the second point: https://mikelevins.github.io/posts/2020-12-18-repl-driven/ https://mikelevins.github.io/posts/2020-12-18-repl-driven/ There might be more CL libraries than you think: https://github.com/CodyReichert/awesome-cl https://github.com/CodyReichert/awesome-cl but I'm not that a zealot to say you can ignore Python's ecosystem, especially in certain areas… [1]: https://www.ravenpack.com/ https://www.ravenpack.com/

catchclose8919 4y ago
> If we can do that in Python, I've been in my cave for too long^^ ...practically speaking you're kind of right, you (almost) can't. In theory one can use `sys.excepthook` to start the python interactive debugger on error and/or %debug jupyter magic, but the "change it, recompile it, come back to the debugger" is not really possible unless you the yuuuuucky thing of changing it from the interactive debugger itself (think change def in code, paster changed def in interactive debugger). As a python expert you can do it. Even modify classes at runtime etc. Tried at various times doing the various things mentioned above. But tbh you learn to rarely do it because you easily get in a twilight zone hell where it's impossible to reproduce anything, no idea what state of code is "active" and what your functions and variables actually are :) ...but basically what Python lacks to enable proper implementations of such feature is decent concurrency - while a thread is paused at the breakloop, the interpreter should allow another to eg. execute code that updates functions and classes definitions - all Jupyter's magic can't do this bc concurrency in the Python interpreter is not doable :) ...in theory the Julia jupyter kernel for example could actually implement all the features mentioned by people describing CL loops - just that someone would need to put the tons of work in, and unfortunately Julia folks seem focused on the automagic-reactive approach to recreating excell-but-as-a-notebook (I don't like this magic-reactive approach tbh... try doing that and have a random variable change "accidentally" triggering recomputing something with a 10gb dataframe you had in var or the retraining of a big deep learning model).

catchclose8919 4y ago

> If we can do that in Python, I've been in my cave for too long^^ ...practically speaking you're kind of right, you (almost) can't. In theory one can use `sys.excepthook` to start the python interactive debugger on error and/or %debug jupyter magic, but the "change it, recompile it, come back to the debugger" is not really possible unless you the yuuuuucky thing of changing it from the interactive debugger itself (think change def in code, paster changed def in interactive debugger). As a python expert you can do it. Even modify classes at runtime etc. Tried at various times doing the various things mentioned above. But tbh you learn to rarely do it because you easily get in a twilight zone hell where it's impossible to reproduce anything, no idea what state of code is "active" and what your functions and variables actually are :) ...but basically what Python lacks to enable proper implementations of such feature is decent concurrency - while a thread is paused at the breakloop, the interpreter should allow another to eg. execute code that updates functions and classes definitions - all Jupyter's magic can't do this bc concurrency in the Python interpreter is not doable :) ...in theory the Julia jupyter kernel for example could actually implement all the features mentioned by people describing CL loops - just that someone would need to put the tons of work in, and unfortunately Julia folks seem focused on the automagic-reactive approach to recreating excell-but-as-a-notebook (I don't like this magic-reactive approach tbh... try doing that and have a random variable change "accidentally" triggering recomputing something with a 10gb dataframe you had in var or the retraining of a big deep learning model).