Hacker News Reader

Comment by tda | Hacker News Reader

Good to know: pycopy is a fork of micropython. The maintainer of pycopy was the second most contributor to micropython, but at somepoint he decided to fork micropython as there were some disagreements.

As far as I'm aware pycopy has a better async implementation. Micropython probably has a bit wider adoption. A lot of libraries are compatible with both, but ymmv

Another notable fork of Micropython is Circuitpython which is maintained by Adafruit.

View on HN

It used to be true that pycopy had some minor improvements over MicroPython's asyncio module. But, recently, the MicroPython asyncio module has been completely overhauled, most of the details are captured here:

https://github.com/micropython/micropython/pull/5332

(Though there have been some additional performance improvements since that was merged.)

The improvements to the old implementation are significant.

I am no longer interested in following pycopy's progress but I'd be shocked if the asyncio implementation was better. Happy to see measurements to suggest otherwise...

> pycopy had some minor improvements over MicroPython's asyncio module

Well, I was the author of "uasyncio" module, as was used with MicroPython. When I switched to Pycopy, I took that module with me.

> The improvements to the old implementation are significant.

Well, how to put it. Before I wrote uasyncio, I wrote asyncio_slow. And before I wrote that, I watched dozens of people writing their own async frameworks for Python. And I watch dozens of people doing the same as we speak. The reason I embarked to write my piece is because I wasn't satisfied with how other people do it. I may imagine you feel the same. So, good luck. (If you do it right, I'll use your stuff.)

> but I'd be shocked if the asyncio implementation was better

It's better, where "better" defined as "more minimalist". If you're not interested in minimalism, then you as well can use CPython's asyncio (and CPython itself).

I haven't followed the recent developments, last time I managed to compile pycopy was end of last year. Would be nice to read about a recent comparison between the two. Is there any reason to prefer one over the other?

Just reading from that PR micropython now also runs picoweb, which is an async webframework that runs on an esp32 and I believe even on an esp8266

If your platform is too small for even pycopy/micropython, there is always snek:

https://sneklang.org/doc/snek.html

Or, you know, C.

I wish ada was more available for microcontrollers.

FreePascal compiles for more platforms than I'd expect, even the Gameboy. Might interest you? I know Ada and Pascal are sort of relatives.

I once targeted ARM/Win32 with freepascal, and was surprised that it got compiled. It even loaded the com objects necessary to build platform user interfaces (barcode reader stuff and showing menus). The problem was, it had dozens of subtle bugs like not bubbling up certain exceptions, maths rounding errors and I simply didn't have the time to fix or report them so I gave up. It probably got better over time but the experience stuck with me and I don't recommend the tool for professional use.

Fair enough, I only use it for basic things on the desktop. I'm not fluent enough in Pascal, and it seems like there's not enough resources out there to make my life easier. Plus nobody pays me to use it, so it's low on my list of tools.

Although not free beer, Mikroelektronic keeps selling their Pascal and Basic compilers.

Or even better some small lisp.

That site is a horrid experience on my Android :(

https://sneklang.org/ is probably better.

Yeah its pretty bad. Likw all the words are one word per line.

iOS too.

I’ve done some RPi projects with Circuitpython and it was a breeze to work with my sensors. Interesting to know it was a fork of Micropython.

Reply on news.ycombinator.com