Conversation

Notices

1. Embed this notice
   David Smith (catfish_man@mastodon.social)'s status on Tuesday, 25-Feb-2025 09:09:39 JST David Smith

   • MainActor

   Hot take: If you really want to make an effective Swift linter just grep for

   * .detached
   * Task { @MainActor
   * any

   and ask "are you sure about that?"

   • Embed this notice
     David Smith (catfish_man@mastodon.social)'s status on Tuesday, 25-Feb-2025 09:09:38 JST David Smith

     in reply to

     • John Siracusa

     @siracusa I actually really disagree! I think assertions are great, and ! is just a compact spelling of one.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 25-Feb-2025 09:09:38 JST Paul Cantrell

     in reply to

     • John Siracusa

     @Catfish_Man @siracusa
     I…kind of agree with both of you?

     Assertions are great and ! is a lovely way to express fail-fast intent •and• in huge swaths of code in the wild, ! just means “I didn’t think about the semantic content of this thing’s optionality, I just made it compile.”

     So I guess write a linter that looks for lines of code where the developer wasn’t thinking, problem solved.

   • Embed this notice
     John Siracusa (siracusa@mastodon.social)'s status on Tuesday, 25-Feb-2025 09:09:39 JST John Siracusa

     in reply to

     @Catfish_Man I’d add: !

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 25-Feb-2025 09:19:35 JST Paul Cantrell

     in reply to

     • John Siracusa

     @Catfish_Man @siracusa
     New idea:

     - Linter that cross-references commit history with target date + dates of especially tense meetings, flags code written under pressure

     Or maybe just:

     - Linter that flags all detected code

   • Embed this notice
     David Smith (catfish_man@mastodon.social)'s status on Tuesday, 25-Feb-2025 09:19:36 JST David Smith

     in reply to

     • Paul Cantrell
     • John Siracusa

     @inthehands @siracusa “how do we get people to stop and think about what they’re doing” remains the greatest unsolved language design question

     Maybe the answer has something to do with reducing the amount of pressure people are under to work quickly… 🙃

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 25-Feb-2025 09:26:57 JST Paul Cantrell

     in reply to

     • John Siracusa

     @Catfish_Man @siracusa
     [tapping head meme]

     “You can’t have code quality problems if you don’t have any computers”

   • Embed this notice
     David Smith (catfish_man@mastodon.social)'s status on Tuesday, 25-Feb-2025 09:26:58 JST David Smith

     in reply to

     • Paul Cantrell
     • John Siracusa

     @inthehands @siracusa

     2029: GitHub introduces a super intelligent AI with a reward function tied to code quality

     2030: global revolution after the AI manipulates news, markets, and the contents of everyone’s GitHub repos to achieve its goal of removing time-to-market pressure

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 25-Feb-2025 09:41:10 JST Paul Cantrell

     in reply to

     • John Siracusa

     @Catfish_Man @siracusa
     After letting this post simmer for a few min, a new grandiose thesis:

     All programming language design decisions boil down to either (1) “more systematized redundancy will finally get programmers to think” or (2) “less redundancy will finally get programmers to think.”