0ch4rl3s.nfo

In light of all the recent decisions made, I've decided to wind down activity here now. I can't use this platform in good conscience, and will be moving to Mastodon. You can find me at @ocharles@mastodon.social. Peace ✌️ @ocharles" target="_blank" rel="nofollow noopener">mastodon.social/@ocharles

@krisajenkins Can you answer 'id'?

My son’s computing homework this week. Right away I see a problem.

A new Haskell coverage report tool is ready to try out! github.com/NorfairKing/de… If you've been looking forward to try it, please have a look. I would be happy with any feedback.

@aaronmblevin I see Source Direct, I give a like. Golden years of drum and bass

a surprise to no one familiar with the people involved, but the Source Direct 12” on Good Looking Records is a real crusher youtu.be/Cqct4H95xQg

@domenkozar NixOS secure boat

NixOS secure boot is being finished on the catamaran #oceansprint

Write your tests purely in terms of the public API, you will thank yourself 😭

Tout plaquer et déposer son CV pour être testeur de jeux de parc pour enfants, clairement c'est l'objectif.

@mpeytonjones @typed_hole This might sound like I'm just moving the goal posts, but there are plenty of cases where there isn't an obvious API/library that you can extract. In that case I keep it internal

@mpeytonjones @typed_hole A user does care - the compiler! I don't think in terms of only the end-user of the whole product. A compiler is really acting like a controller, delegating work to subcomponents and then aggregating - it really doesn't do a huge amount. foo-ir sounds like a good library.

@mpeytonjones @typed_hole I think exposing sub-parts can lead to what A Philosophy of Software Design calls "temporal leakage" (5.3)

@mpeytonjones @typed_hole My personal take is biased to the first point though. If testing this composite becomes too difficult I take a step back and work out what change would make testing easier. Exposing all sub-parts is usually the answer answer, but not necessarily the best one.

@kerckhove_ts @typed_hole Exactly, it's fine to be your own user!

@acid2 @typed_hole You hit the nail on the head with this. The question is not whether you test private functions. The question is "who is the user?".

@typed_hole If you truly need to test something private that's probably a signal that there's a little library with a good API hiding in there

@typed_hole 💯 I never got the argument that we should be able to test private functions. No, test the things your users actually get to use!

@kosmikus @tomjaguarpaw Because it's not at all obvious what the intent of the call is. It's like using `id` for `traverse`. That said, perhaps there's a contribution of `traverseLens = toLensVL`?

@acid2 @tomjaguarpaw I don't quite understand why toLensVL is unsatisfying.

I'm exploring replacing a project's use of "lens" with "optics". Does anyone know how I traverse over a lens (not a traversal)? In lens this is `traverseOf`, but `traverseOf` in optics requires `Applicative f` (but `Functor f` is enough).

@tomjaguarpaw Yea, that's all I found so far

@acid2 I guess toLensVL is the unsatisfying answer: #v:toLensVL" target="_blank" rel="nofollow noopener">hackage.haskell.org/package/optics…

@Profpatsch Oh wait, almost two years! No, I don't believe that.

@Profpatsch Almost a year!

Days since I lost hours of producitivity due accidentally writing let x = x: 0