@willmcgugan Fully agreed, and Mexican-friends approved. My team liked it so much when I invited them that I got everyone one
English
Marco Weber
221 posts
@haxor Some syntax highlighters (is that a word?) use capital R vs lowercase r to distinguish regexes from raw strings
English
Today I learned that Python supports 100 ways to write a string literal. How do you decide which to use?
4 quote types:
' " ''' """
25 prefixes:
(None) r R b B u U f F br BR bR BR rb RB Rb RB fr Fr fR FR rf rF Rf RF
These are the 20 variations I'd say are worth knowing:
English
@willmcgugan The fix should only work for Python 3.11 and above, right? Because only since then time.sleep uses the high resolution timer
English
A better asyncio sleep for Windows, to fix animation in Textual
textual.textualize.io/blog/2022/12/3…
English
@willmcgugan @MarximusMaximus How did you solve this? I had the same issue with a network packet generator that deviated too much from the desired data rate. In the end I added kind of a control control that measured the actual sleep interval with the high resolution timer
English
@MarximusMaximus It was the granularity of Textual's Timer class. Which depends on asycio.sleep
It looks like asyncio.sleep has a granularity of 15ms on Windows. i.e. you sleep 16ms and it actually wakes up after 30ms
English
Oh good $DEITY finally!
I've struggled to get decent performance with Textual on Windows. Nothing seemed to make any difference.
I assumed Windows Terminal was just slow to render update. Which seemed implausible given the amount of effort that Microsoft was putting in to it.
English
@dontusethiscode Some people will see all the red fields as a challenge
English
I know at least three ways to mutate a tuple in Python…
Cameron Riddell@RiddleMeCam
I write blog posts featured in our weekly newsletter. This weeks post will discuss common misconceptions about the relationship between hashability and mutability in Python. Sign up now to take a peek!
English
@ID_AA_Carmack If IQ tests and tech Twitter had a child: "TensorFlow behaves to PyTorch like gcc to llvm"
English
Some will probably hate the characterization, but TensorFlow feels like gcc to PyTorch and llvm — still critical infrastructure that will be around for decades, but no longer ascendant.
English
@willmcgugan A shame "poor" seems to have been namespace squatted on pypi, otherwise you could have used it for your parameterless version of rich
English
@pyblogsal Also love those quality of life improvements in datetime.fromisoformat and enums (especially str/format behavior)
English
@pyblogsal I thought asyncio.TaskGroups were the "killer feature" of this release, although admittedly they are not relevant to most users.
English
@ChadAChavez @hammer_mt @AlSweigart Probably both. As others have pointed out, Arondir (the black elf) plays a major role and is neither a coward nor a jerk - doesn't he count?
English
@webology @samwillis @pyscript_dev @nedbat The expecthook and restart as a subprocess was more for dramatic effect though. I want to try out something based on module __getattr__, but I still haven't figured how to map install name to import name (in case they differ)
English
@comabrewer @samwillis @pyscript_dev @nedbat `abomination.magic()` looks like it might solve the itch I was looking for.
If the "quoted:version" format doesn't work, I have wondered how hard it'd be to create something in the vein of f-strings or raw-strings format-wise.
English
🤔 About once a month, I wish Python supported a similar syntax:
> import "pypi:typer@0.6"
Wes Bos@wesbos
Interesting - Deno is adding npm package support. Just import it with the `npm:` prefix, and it will download and cache it for you. deno.com/blog/changes
English
@webology @samwillis @pyscript_dev @nedbat It should definitely be possible to create something like "from pypi import typer.v0_6_0", which is valid syntax. Pip normalizes separators in the package name to dashes anyway. The leading v and the underscores should be enough to make the version number acceptable
English
@webology @samwillis @pyscript_dev @nedbat A while ago I hacked together something related, minus the version specification (also inspired by twitter): #L9" target="_blank" rel="nofollow noopener">github.com/comabrewer/abo…
English
@haxor That's why I like pytest's tmp_path fixture. Given that you can use pytest and don't have issues with the fixture mechanism or pathlib's path objects, you often don't need any of those imports
English
import os
import shutil
import tempfile
import unittest
... guess who's writing tests that must interact with the filesystem?
English
@willmcgugan TFW you think the world agrees with you just to realize that you were the first to answer because you are such a snek Twitter addict
English
@HeyChelseaTroy The "what makes this song great" episode by Rick Beato on YouTube has really helped me learn appreciate this song, its incredible production and the perfect guitar solo
English
@HenrySchreiner3 @cs01_software I didn't want to actually suggest a name change, but just make a cheap (half)joke. As a C++ developer where I must, Python developer where I can I'm really excited about your past, present and future work, under whatever name it may be
English
@comabrewer @cs01_software Probably; also scikit-build is nearly 10 years old, so not sure renaming it was really on the table. :) The name scikit-build-core for the backend might be something that is not yet set in stone.
English
FYI, I’ve let the cat out of the bag today at SciPy, so will share here too: iscinumpy.dev/post/scikit-bu… has been accepted - I’ve been funded to work on Scikit-build for the next three years. Expect great things for binary package building! :D
English
@cs01_software @HenrySchreiner3 Just guessing, but it could have been more difficult to secure funding for "cmake-build" 😁
English
@HenrySchreiner3 Congrats, very exciting! I read the link, and I’m curious why it is called scikit-build rather than cmake-build or something along those lines. The scope seems broader than scientific only.
English
@brianokken Sometimes I write that tiny program directly as a test (with pytest of course) - I write a test that makes assertions not about my own code, but about the library's code
English
@brettsky Wait, you can just add a .gitignore *inside* the folder to be ignored so you don't have to commit anything? Why have I never thought of this
English