It seems to be a normal cabal multi-package project. As evidenced by cabal.project.

Looking at the Nix, it uses haskell.nix which itself runs the solver etc. So I wouldn’t be surprised if you could run cabal-install on this project without Nix’s involvement.

I would like to talk about the recently announced tricorder, since it is exactly a “tool for leveraging LLMs”!

Hey everyone,

for those of you who don’t know me, my name is Christian and I’m an engineer at Tweag.

I am very pleased to announce the very first release of tricorder!

tricorder has been my daily driver for Haskell development for months now. It works really well as a standalone replacement for ghcid or ghciwatch with some features that make it very attractive:

• Multiple component…

I’m surprised/disappointed. This gets rejected:

t42 = Typed (type Int) 42

    Unexpected keyword `type’
    Suggested fix: … `ExplicitNamespaces’ … (implied by …)

If there’s anything that implies that extension, surely it would be RequiredTypeArguments(?) I particularly remember that stipulation in the discussion.

• The flip situation is indeed rather disappointing. For reference, Agda’s type for flip looks like this:

  flip : ∀ {A : Set a} {B : Set b} {C : A → B → Set c} →
         ((x : A) (y : B) → C x y) → ((y : B) (x : A) → C x y)
  flip f = λ y x → f x y
  

And it relies on higher-order unification to instantiate C at use sites. It’s not clear to me what the Haskell solution…

`fromText . Data.Text.singleton` will do it for you, using fromText.

in ghci :
- I played the file (i.e. saved in music.hs) using my piano (BlueTooth MIDI #7),

• saved to a midi file called music.midi and

• opened the file using MuseScore

  Loaded package environment from /Users/me/.ghc/aarch64-darwin-9.6.6/environments/default
  GHCi, version 9.6.6:
  :? for help
  ghci> :l music.hs
  [1 of 2] Compiling Main ( music.hs,…

unsafeFromChar :: Char → OsChar

maybe define a short function alias `och` (also funny if you’re german)

Hi all!

_tl;dr: I have 148 packages on stackage to patch for compatibility, and I’d like your help!_

I’ve got another update on the Monad of No Return/Monoid of No Mappend proposal.

Following proposal acceptance just under a year ago, I started implementation this year. Recently the MR is in a good enough state for me to check it against clc-stackage, a tool for building all of stackage…

unsafeFromChar isn’t equivalent to osp + head, if i give it a invalid literal i will get a compile time error, unsafeFromChar will just fail silently as it truncates the bytes. I guess if i give osp the empty string then it also fails at runtime but it atleast will raise an exception

I can certainly add such a quasiquoter if you find it useful.

Feel free to create a PR. The osstr quasiquoter is here for an example: os-string/System/OsString/Internal.hs at 0cc9aded7e582c508fb4246700aaecf7956d9877 · haskell/os-string · GitHub

i started trying to use OsString + OsChar and immediately ran into this issue using osp is really annoying but tolerable this is more than double the overhead

Ah, I understand what you mean now! Yes, the last-one ends up winning. I think I’ll move to a cleaner high level interface (probably something more like setPublic etc) that would make this behaviour more obvious and harder to accidentally trip up on.

Thanks for the info. I was mostly wondering about a case such as marking something as both public and private—ideally that wouldn’t be possible, but maybe it’s last-one-wins?

Alternatively, if it were an argument to runClassBuilder then it would be syntactically impossible to supply multiple, but then it would also be a required parameter (no defaulting).

Not necessarily a big problem, it’s…

Indeed—it’s a tree of function compositions, but that’s the implementation, not the exposed interface. So my thoughts are around whether there’s a better way to implement that interface.

@evincar wondering your thoughts (and anyone else’s of course!) on making the MonadFix approach the only way to emit labels. I think this would make the UnmarkedLabel error mostly impossible (apart from weird cases like sharing labels across CodeBuilders, which could be prevented with a state variable like ST does). However it also forces a certain style upon users, which might be a bit…

Haskell Weekly

### Issue 527

News about the Haskell programming language from 2026-06-04.

It’s a fun game! Thanks for sharing. I thought I’d easily score 12/12 but I totally fell for the gotchas!

Thank you for the feedback!

All addAccessFlag does is prepends it to a list in the underlying state monad, so nothing super interesting. You’re able to use it with multiple different flags in the way you’d expect. I’ll consider adding an overload to add multiple at once.

There’s no check for duplicates, but since the flags compile down to a simple bitmask, nothing weird would happen if you…

Wow, this is all amazing, thank you so much for such thorough feedback!

The MonadFix suggestion is fantastic, I’m not too familiar with that part of the language but this does seem like a perfect usage!

I’ll try and find a more elegant way of expressing the reversed list, as I see there’s no single consensus about the best option. At the moment I’m probably leaning towards a simple newtype…

-- Expressions
t1 = Typed Int 42
t2 = Typed String “hello”
t3 = Typed (Int → Bool) even

I’m so used to putting the type following the expr, this still doesn’t feel right (even though I knew it was coming). Is it really not possible to support

t1’ = Typed 42 (type Int)

  Typed :: forall a. a → (type a) → T a

jonathanknowles:

Even though the parsec-3.1.18.0 boot package bundled with GHC 9.14.1 was itself compiled against base-4.22.0.0, the version on Hackage (with an outdated upper bound on base) seems to cause dependency resolution failures for packages that transitively depend on parsec-3.1.18.0 when building with GHC 9.14.1.

This looks like an error in the GHC release…

Thank you Vladislav.

Good to hear you could rescue the StarIsType syntax. TIL (*) → (*) is also a valid kind. That looks even more like an operator section.

Might kind signatures turn up in patterns in future work? Is there a risk of confusing the kind -> with a ViewPattern?

There’s a GHC plugin that compiles all the mutations into the same binary so that you still only need to compile once. The mutations are then activated one-by-one at runtime by the mutation engine.

I wrote this little Haskell Quiz (with help from some friends); in the style of “CSS or BS”; Can you tell the valid Haskell Language Extensions (based on the GHC docs), from the Impostors?

doscienceto.it

Extension or Imitation?

Can you tell the real Haskell Language Extensions from the Imitations?

it’s a tree of function compositions

Sure; DList can do more, but for the common case of just needing to accumulate a list in the opposite order of the desired order of the result, it seems like overkill (both in terms of intelligibility and allocation).

I feel like people get the impression that using function composition is somehow leading to some aspect having zero overhead, but it isn’t. And in particular, the final toList…

But it can also store the list in forward order. You can both cons and snoc

not a FOSS license tho..

evincar:

instead of storing a list in reverse order you can use a DList

I’m always suspicious of DList in that physically it is essentially storing a list in reverse order, just less clearly: each closure is closing over one new element (actually, over another closure capturing that element) and over the closure recursively referencing the rest of the elements, which is the same structure…