Simon Hengel

Obsessed with software correctness.

[Simon Hengel @solirc_]

...the single biggest benefit of unit testing is that it leads to a more well-factored, more functional and more pure code base!

[Simon Hengel @solirc_]

What is easier in 2024, creating and maintaining a neovim configuration, or just implement an editor from scratch and be done with it?

[Simon Hengel @solirc_]

@kerckhove_ts https://t.co/FSyCdL87sQ?

[Simon Hengel @solirc_]

@ChShersh That's called lean development.

[Simon Hengel @solirc_]

@vamchale Not Jacinda though, at least not yet 😅

[Simon Hengel @solirc_]

@vamchale Take this a step further and automatically tag and release to Hackage on version bumps: https://t.co/Iz5B9lTZLv

[Simon Hengel @solirc_]

@vamchale Here is what this does:
1, If there there is not yet a git tag for the version in the .cabal file then create it.
2. If a new tag has been created then also publish to Hackage.

[Simon Hengel @solirc_]

@ChShersh @etrepum That's exactly the issue here. You would need tests to verify that this list comprehension does what you expect it to do, kind of defeating the whole purpose of this endeavor.

[Simon Hengel @solirc_]

With home row mods you barely need to reach for ← BACKSPACE anymore.

Use CTRL-H (on the terminal) instead!

For Dvorak it's index home, for QWERTY, Colemak and Colemak-DH it's index ODFH 🎉🎉🎉

[Simon Hengel @solirc_]

@oshwassociation https://t.co/9TQMMGhCyl seems to be down. Is there some other space to discuss all things open source hardware?

[Simon Hengel @solirc_]

@ChShersh Yes, and yes... not really a good solution, but I tend to standalone derive instances when needed.. also in GHCi try :eval and/or :force...

[Simon Hengel @solirc_]

@mattoflambda What are you aiming at? Eager to see some code.

[Matt Parsons @mattoflambda]

Hspec Hooks https://t.co/fivfcHgdR9

[Simon Hengel @solirc_]

@s_anastasov @jdegoes 2/2

What I'm saying is:

1. In languages like Haskell we have to do it explicit, because we can't rely on reflection.

2. We do it all the time. It's so common that we don't think about it and we don't give it a fancy name.

[Simon Hengel @solirc_]

@s_anastasov @jdegoes 1/2

If I have some piece of code that e.g. uses `getTime :: IO Time` and rather than using it directly I'm passing it in as a dependency (read function), isn't that conceptually the same thing as DI?

[Simon Hengel @solirc_]

@GabriellaG439 😆

[Simon Hengel @solirc_]

@ChShersh @EyalL And this propagates transitively to the 955 direct + who knows how many indirect dependencies.

And no, this hasn't been always like this, it's a recent "improvement" of `hashable-1.4.0.0`.

#just_a_bunch_of_kids_playing_with_their_toys

[Simon Hengel @solirc_]

What's the idea behind `hashable` being a separate package from `unordered-containers` AND MORE IMPORTANTLY what's wrong with this picture? (asking for a friend)

[Simon Hengel @solirc_]

@ChShersh @EyalL For `base-orphans` that's actually worse. It doesn't (yet) matter, as only the latest version of GHC is excluded. But imagine the situation when GHC 9.4.* is released and any new instances are added to `base-orphans`. Now GHC < 9.2 will have these and > 9.2 too, but not 9.2.

[Simon Hengel @solirc_]

@mattoflambda Maybe scratch that newtype and just add an orphan `Example` instance for `SqlPersistT IO ()` or something.