Drop Python bindings? about libmonome

[artfwo]

@artfwo I assume that most people would/should be using pymonome instead?

generally, yes, reasons are: there's a tutorial on grid studies, the API is a bit more pythonic, it uses the OSC protocol and so handles multiple devices gracefully.

Are there any advantages to libmonome's Python bindings over pymonome?

It doesn't need serialosc, but then again, i don't know if that's an advantage - the rest of "bindings" we have on grid studies also depend on serialosc.

For extreme cases you can always use pyserial with standardised event loops, which I think I would prefer over libmonome's module.

[tehn]

i'd be fine dropping it, but i'm also not aware if anyone is actually using it. as far as i know @artfwo is the foremost monome-in-python user :)

[simonvanderveldt]

@artfwo Do you think there would be an advantage in refresh rate for libmonome's Python bindings over pymonome?

[artfwo]

@simonvanderveldt when directly accessing devices through libmonome you don't hit the the UDP buffer size limits (which you will hit with serialosc), you will hit the serial i/o buffer though. in practice the framerates in both cases can go crazy high (beyond visible refreshing), for example i currently run the arc at 100 fps through serialosc/pymonome in my live setup, but if I set the framerate to 1000 it will still work - actual framerate will be probably lower due to some UDP packets skipped, but i won't be able to notice this with naked eye.

[simonvanderveldt]

@artfwo Thanks for the explanation, sounds like pymonome is plenty fast :)
Unless anything else comes up it seems like it should be fine to drop the Python bindings I guess? Shall I try to create a PR or would you prefer that you do it?

[artfwo]

@simonvanderveldt please do so if you like, sorry, i have somehow busy week again.

[catfact]

i'm gonna go ahead and nuke the python bindings and monomeserial from this library, which will henceforth be focused on offering a clean C binding to serial devices. (since the other tools handle the other jobs better, and are better maintained.)

[catfact]

on second thought, may as well leave the Cython bindings. it does offer a lightweight means of directly controlling the serial device, handling protocol detection/rotation/etc.

[artfwo]

@catfact an alternative to Cython bindings is pyserial that provides a more pythonic API and is compatible with standard Python event loops (pyserial-asyncio), but the downside is of course that you have to write a grid protocol layer on top of it.

[shakfu]

Please don't nuke the python bindings:

1. It works works perfectly with the latest python (tested on 3.9.9)
2. The cython code is really minimal (without the header section which doesn't change, it's just 861 words (using wc -w) and 343 lines (using wc -l) which is hardly a maintenance burden.
3. It's a great place for python-centric people to grok the c-api and experiment with libmonome via python scripting.