MicroPython Forum (Archive)

So, based on the analysis above, I can propose different ways to solve it: 1. [...] 4. We change function signature of .add_reader() and friends - instead of accepting "file descriptor", as CPython asyncio mandates, it would accept "file-like object", and .add_reader() can decide what to do with it...

pfalcon wrote:For completeness, here's testcase which is almost 10,000 times slower on CPython:

Interesting. But change "utf-8" to something else and pass sth other than 0-bytes and Micropython will give wrong results.

That would also explain the performance improvement of pad3 over pad2. len(s) seems relatively costly.

But AFAIK Python 3.5 uses a variable-byte internal string representation now as well (but not utf-8).

Beyond that, your test contain logical error which of course affect results. String is a sequence of characters. It makes no sense to pad it for 4-bytes alignment - you cannot assume anything about "byte size" of characters. The original functions I based my test script on were receiving a str and ...

I was profiling some different functions for padding binary strings with zeros for a 4-byte alignment and I noticed that micropython (unix) performs really badly compared to CPython (let alone PyPy) at this. For comparison, I also timed a simple numerical function (fibonacci) and you can see that mi...

pfalcon wrote:bytes is immutable object, so it's only suitable for read-only arguments, which won't be changed. bytearray is mutable, and is suitable for buffers which will be written to.

This seems to tell otherwise:

https://github.com/micropython/micropyt ... /re.py#L68

So, I eventually figured out inet_ntop : import socket import struct INET_ADDRSTRLEN = 16 inet_ntop = getattr(socket, 'inet_ntop', None) if not inet_ntop: import ffilib _inet_ntop = ffilib.libc().func("s", "inet_ntop", "iPpi") def inet_ntop(a): buf = bytearray(INET_ADDRSTRLEN) res = _inet_ntop(socke...

pfalcon wrote:If you know Python, it should be pretty clear, that you use "bytes" for input (read only) parameter, and bytearray for output (read/write) parameter.

Sorry, I program Python for almost 20 years, but I don't see how that's "pretty clear". To a C programmer, maybe.

Your C expertise doesn't let you read comments in English: https://github.com/micropython/micropython/blob/master/unix/modffi.c#L39 ? It actually is sufficient to infer most of the information in this comment* from https://github.com/micropython/micropython/blob/master/unix/modffi.c#L101, but not t...

The first place to look at should be a source of a module. I tried, but it's a bit past my C expertise. And ffi module specifically is used a lot in micropython-lib, you can look there for various usage examples. Allright, it didn't occur to me to look there :( So, I always was under the impression...