Conversation

Notices

1. Embed this notice
   ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 21-Apr-2025 08:42:31 JST ✧✦Catherine✦✧

   python 3.12 match statement makes it really easy to write ad-hoc option parsers with state machines

   • clacke and clacke at libranet.de is my main like this.
   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 21-Apr-2025 08:50:11 JST ✧✦Catherine✦✧

     in reply to

     yes yes i fucked up. this needs a while loop or something

   • Embed this notice
     Helge Heß (helge@mastodon.social)'s status on Monday, 21-Apr-2025 08:58:44 JST Helge Heß

     in reply to

     @whitequark Looks like Swift, except in Swift, you rather link a 6MB big argument-parser lib that statically types out all the possible options 🙈
     (but really, the `match` looks like how Swift's `switch` works)

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 21-Apr-2025 09:06:00 JST ✧✦Catherine✦✧

     in reply to

     • Simon Sapin

     @simon ... excellent idea

   • Embed this notice
     Simon Sapin (simon@tutut.delire.party)'s status on Monday, 21-Apr-2025 09:06:02 JST Simon Sapin

     in reply to

     @whitequark or name the iterator and call next() on it for --triple?

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 21-Apr-2025 09:06:15 JST ✧✦Catherine✦✧

     in reply to

     • Simon Sapin

     @simon actually no that won't work because you need lookahead

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 21-Apr-2025 09:06:56 JST ✧✦Catherine✦✧

     in reply to

     • Helge Heß

     @helge or Rust! yeah

   • Embed this notice
     poleguy (poleguy@mastodon.social)'s status on Monday, 21-Apr-2025 11:52:17 JST poleguy

     in reply to

     @whitequark I just use typer. Good enough.

   • Embed this notice
     dram🎀 (dramforever@mastodon.social)'s status on Sunday, 27-Apr-2025 02:19:18 JST dram🎀

     in reply to

     @whitequark i had an idea... and yup

     while True:
     match args:
     case [ '--foo', foo, *args ]:
     print('foo', foo)
     case [ '--baz', *args ]:
     print('baz')
     case [ '--', *args ]:
     break
     case [ arg, *args ] if arg.startswith('-'):
     print('Unrecognized', arg)
     raise RuntimeError() # FIXME

     # This can't be case _:
     # Otherwise previous case overwrites args
     case args:
     break

     clacke likes this.
   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Sunday, 27-Apr-2025 05:37:18 JST ✧✦Catherine✦✧

     in reply to

     • dram🎀

     @dramforever oh this is cute

   • Embed this notice
     dram🎀 (dramforever@mastodon.social)'s status on Sunday, 27-Apr-2025 05:37:19 JST dram🎀

     in reply to

     @whitequark now i wonder if you can subclass list to make it not quadratic

   • Embed this notice
     Simon Sapin (simon@tutut.delire.party)'s status on Sunday, 27-Apr-2025 16:00:12 JST Simon Sapin

     in reply to

     @whitequark wow i didn’t know modern python could be this nice
     (shouldn’t `--triple something` consume two arguments though?)

     clacke likes this.
   • Embed this notice
     Thomas Guyot-Sionnest (dermoth@noc.social)'s status on Sunday, 27-Apr-2025 16:00:21 JST Thomas Guyot-Sionnest

     in reply to

     @whitequark What I really like is matching directly object or property types in the case statements. ex:

     ```python
     match fruit_obj:
     case Apple(color=Red()):
     print("I'm a red apple")
     case Apple():
     print("I'm an apple")
     case Orange():
     print("I'm an orange")
     case _:
     print(f"Unknown fruit type: {type(fruit_obj)}")
     ```

     P.S. match case statements were introduced in 3.10.

     clacke likes this.