Guaranteed Safety Through Effect and Capability Systems
I think effects would be natural interfaces for privilege separated components actually.
Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.
I think effects would be natural interfaces for privilege separated components actually.
ByteString is a ForeignPtr Word8 under the hood, and I then wondered why withForeignPtr needs to run in IO if it’s just peeking into memory positions. Ptr and ForeignPtr are both mutable, so any reading must happen in IO. My solution was to write my own PinnedByteArray type, using ByteArray# under the hood, and change networking and parsing code to use that instead. You could use the unboxed variant of one of the off-the-shelf array libraries, like vector, array, etc. (I found all those to be a…
Open issue haskell/core-libraries-committee Fix `System.IO.hGetContents'` to close the handle on exception opened 12:11PM - 16 Nov 26 UTC Lysxia `hGetContents'` reads all of the input on a handle. When it completes normally, …it naturally closes the handle. But when it encounters a decoding error, it leaves the handle open. This is unlike the lazy `hGetContents` which closes the handle on error. As the one who wrote `hGetContents'` I can say this behavior was not intentional, just an oversight.…
Open issue haskell/core-libraries-committee Safe access to carry-type functions from `GHC.Exts` opened 07:58PM - 08 Oct 26 UTC Zemyla # Motivation If I want to write a function that checks for overflow when adding …or multiplying `Word`-like values, I have to either write it myself (and possibly make mistakes), or import `GHC.Exts` and still write the wrapper myself (which will soon depend on `ghc-internal`). I feel like this is silly when the primops exist. # Proposal Create a new typeclass…
This is very interesting, and something I’ve wanted in order to run Bluefin code in GHCi. I guess n has to have a Monad instance, and each command in the nested block must be of type n. Is that right? I don’t see how to make that work with Bluefin, where the functions I’d be working with have types like (forall e. Throw ex e -> Eff (e :& es) r) -> Eff es r, that is, the nested block has something new in scope (the Throw ex e). Can that be accommodated in your setup? (Of course, we could use…
This is very interesting write-up! I am writing my own PostgreSQL pure Haskell client library (and the reasons why are for the whole other topic that I will start once in alpha ). Among other more or less unorthodox decisions, I moved parser composition to type level using typeclass metaprogramming - it lets GHC optimize the whole thing to tight non-allocating loops.

[ANN] Brillo 3.0.0, a simple GLFW-OpenGL based graphics library is released Announcements we’ve cleared up some gloss technical debt for this version lol, changelog: Breaking: Add a Shader constructor to Picture for rendering custom GLSL fragment shaders on a quad (compiled lazily and cached by source text) Add UniformValue and ShaderData types, a shader convenience function, and defaultShaderVertex Add brillo-shader example with rounded rectangle and flame shaders Breaking: Report mouse wheel…
@Execute told you we ought to have a marketing dept early!

Never considered my package updates as an emergency Maybe we should have an [UPD]ates section for that instead. The examples compare 3 incremental updates for some old stuff with 3 new things. What I personally would appreciate is Hackage showing the package summary on the changelog links, like it does on the regular links already. And more people linking changelogs. Also it is really awkward having to cram everything into a title. But maybe that gives better publicity at least.

@Execute It depends from the target audience! If you wrote your ANN to just users who already use you library - then you are right - no repetative “water” is needed in the description. But… if you also wish to advertise your library to whoom it could be interested in - then, you are not right! You must also advertize - that means describe about what your lib is for and why it is better then other libs in the same Announce!
This happens when your account gets deactivated for inactivity. Please DM (!) me your handle+mail address and I’ll reactivate your account. Not just to you: I strongly recommend to not to try and “remember” passwords and instead use a password manager. Remembering a 20+ chars strings of random symbols is very hard for humans (and you shouldn’t use your cat’s birthday). Reason I’m saying this is that I have resolved this exact issue already 10+ times in the last year and that’s just the people…
Oh yeah totally. The main current limitation that I feel compelled to warn about is that the engine doesn’t have any official physics system yet, meaning you’ll have to roll your own or use bindings for something like box2d.
Very cool! Looking forward to trying it out, it looks very complete Sending regions to the repl is something I’ve missed. Does it work OK with code that uses quasiquotes? E.g. ihp-hsx: JSX-like but for Haskell And should it work with direnv/envrc? curry-mode warns you when the installed grammar is older than the one it expects In my experience with tree-sitter, maybe it should also warn when it’s newer..

I’d like to log in on https://gitlab.haskell.org/ but I’ve forgotten the password. When trying to reset it I don’t receive any email. (No, it doesn’t end up in my spam folder.) I thought maybe my account was deleted since I haven’t been very active, but creating an account with my email address fails due to it already being in use. I’m hoping someone here knows where I should turn to get this sorted.
yeah sorry, I just thought it is pretty redundant to write the same thing twice when people could’ve just clicked the link to see
OK, thanks! You are right. I’ll take that into account and try to make it more transparent when I have time.
Yep, that’s all that’s needed. It’s not a lot of work to make these posts so much more useful to many more people.
so copy and paste the hackage description/tagline so people don’t have to click thru?
I recently found DeepWiki, which provides automatic AI documentation for any public repo, which is super helpful https://deepwiki.com/haskell/cabal
For fast non-interactive non-fuss rendering you can take massiv-io: Import/export of Image files into massiv Arrays It also wraps JuicyPixels but binds it to massiv: “Multi-dimensional Arrays with fusion, stencils and parallel computation”.
The main package got a major bump: Changelog for dear-imgui-3.0.0 | Hackage That means a large rewrite beyond mere breakage. The rewrite in question is dear-imgui-raw: Raw FFI bindings to dear_bindings. The raw bindings now come in their own package, generated from a bunch of dear_bindings json specs and their own C wrappers. Custom C parser is gone, manual inline-c ++ wrappers are gone. Since it costs nothing, the wrappers are binding the docking/multiview branch, which you can opt-in at…

I guess there are two direction since the cabal hooks and buck2 backend Buck2 build system for Cabal projects Announcements This started as a little experiment and got a bit out of hand, but anyway: I’ve built an extension to Cabal to let you use Buck2 as the build system for a Cabal project. Why, might you ask? Well, it’s faster for one thing: [perf-chart] There are two parts to this: cabal buck2, which currently lives in a fork of the cabal repo A modified Buck2 prelude and some support code,…
I’d recommend brillo, that also allows you to animate it (e.g. zooming in) in the future. You can combine it with JuicyPixels using the brillo-juicy package.
Hi all :), Cabal’s future architecture. How would that look like? Couldn’t find a document. Did I overlook something? If so, please point me to it. So, I asked AI to compile one from the bits and pieces out there. Then, to which extent are current efforts (PRs, issues) aligned to that? Find both diagrams here. I feel this might be valuable to further a discussion realising such a future architecture document. Assumption here is if the target is clear, efforts could be better aligned towards…

After private discussions with members of the board and others, Travis Whitaker has decided to resign as a board member, effectively immediately. Directors resigning shortly after their appointment has been announced is not how we would want board elections to proceed, and the board will be reviewing events to seek to avoid any repeats of this situation. As a final note, the Haskell Foundation retains its commitment to representing the entire Haskell community, and to the guidelines for…
For a Mandelbrot set program, I would (personally) reach for JuicyPixels
Calling all Londoners (and people who can get to London)! On Thursday 26th of November, we will be running a Haskell talk at Imperial College London. Doors open at 18:30, the talk starts at 19:00 If you’re interested, sign up to the event! If you’re interested in submitting a talk for a future event, please use the form. If you have any questions, please email us at london.haskell.meetup [AT] gmail.com. Join us on Discord: London Haskell This month’s event is sponsored by Obsidian Systems. The…
Hi all, I’m trying to generate some 2d pictures of the Mandelbrot set (& similars) following along with this paper; however the paper is from 2004 and I’m not sure the graphics library they use still exists or is recommended. is there any easy to use, well maintained library I could use? If possible I’d like to be able to generate high resolution images and export to file (png/jpeg or whatever). Thanks!