What is static analysis, really?
mkleczek: Why Writing Correct Software Is Hard - amazing talk giving a nice overview of the difficulties and limits of static analysis. Oh yeah looks pretty interesting, will take a look, thanks!
Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.
mkleczek: Why Writing Correct Software Is Hard - amazing talk giving a nice overview of the difficulties and limits of static analysis. Oh yeah looks pretty interesting, will take a look, thanks!
Thanks for the brief introduction to the subject - whenever it comes up, my first answer is always this link: Why Writing Correct Software Is Hard - amazing talk giving a nice overview of the difficulties and limits of static analysis.
Jappie: Maybe you don’t know, but in opensource development when people fork a project they’re saying “I can’t work with these maintainers”. You didn’t even try so it’s just strange. You do realize there’s a ton of other potential different reasons for him making a fork? No idea why are you assuming this one.
brillo-shader example (IDK what happened and why it’s glitching at the end)

from his perspective it’s just kind of a bad feeling when u care about ur own project for years and somebody posts a large fork about it without having communicated anything with u beforehand, regardless of licensing or sharing it with the community. and the tone quickly degraded towards the chain of replies just mostly getting a non answer really. regardless of how jappie has phrased it, it’s more about the moral value of having talked about it beforehand, I’m sure jappie would have been happy…
LaurentRDC: Jappie: when people fork a project they’re saying “I can’t work with these maintainers” I don’t think you should see it like that @Jappie, see Fourmolu being a fork of Ormolu because of different views on configuration. Well, it depends on how you interpret that statement. In the case of Fourmolu, it wasn’t personal (Mark seems like a great guy), but it was effectively “I can’t work with these maintainers [because their goals are in direct conflict with mine]”. Anyway, I do agree…
Thank you for writing up that ticket. It’s so much better than I’d have done it.

A first preview is ready! Test the development branch on your machine NOW: Adding loading indicators to the build command · Issue #8943 · haskell/cabal · GitHub (It still needs polishing and refinement, so let me know what comes to your attention while testing.)

What did an LLM do for you today? Hopefully something a lot more valuable than this - but if it raises a smile and makes someone’s day a little more fun then that’s a good use too right? I'm pure, Im lazy
Is anyone working on (or interested in) opt-in totality checking for Haskell I’m not working on it but I definitely agree it would be a cool feature (especially in combination with Dependent Haskell). Of course it would be incomplete because of #halting #problem but modern termination analysis is quite sophisticated and with some added features like an escape hatch or the ability to include a loop invariant I think it would be quite useful. Productivity checking too, of course. There has been…
Ah, thanks for clarifying. I have no idea why I thought that then!
My question is, why Haskell? Why not generate Idris or Lean instead, since you seem to already rely on Lean proofs?
tomjaguarpaw: GHC will only be released once a year now rather than twice a year as it was in the past. 10.0 having taken a year was not entirely by choice. Or more specifically choices to delay the release rather than release with a version we were unhappy with have led to this. And it should not be taken as more than that. Currently to plan is to aim for 6 months after 10.0 rather than another 12 months for 10.2.
Maybe @Jappie wish to tell that it is very easy to fork a project, but it is very hard to maintain this fork fresh and issueless for significant amount of time - even years.
Jappie: when people fork a project they’re saying “I can’t work with these maintainers” I don’t think you should see it like that @Jappie, see Fourmolu being a fork of Ormolu because of different views on configuration. It should be noted that in my experience @Jappie is super open to contributions and ideas.
It would’ve been nice you attempted to communicate, or tried upstreaming changes into mysql-haskell before announcing a fork. Maybe you don’t know, but in opensource development when people fork a project they’re saying “I can’t work with these maintainers”. You didn’t even try so it’s just strange. and no I don’t get paid for any of this although I do take a bit of pride in keeping it functioning all these years and see usage grow, recently I got even some contributions and interest. I suppose…
Dude, its like you are getting paid for it or something. We created something we needed, shared it with the community, whats the big deal?
so instead of filing an issue, you created a whole new package with many thousands of lines of new code and signing yourself up for maintenance in perpetuity, claiming this is saving you time?
Yes, there are benchmarks in github comparing performance on standard mysql. Also, mysql-haskell actually fails 20+ scenarios on Maria, which was one of the primary reasons to create this.
Yes, unfortunately MariaDB and MySQL have diverged quite significantly and the new package uses a lot of native Maria APIs. Making it a module might be possible, but is much more work.
Hi, I’m the maintainer of mysql-haskell, it’s a bit strange to fork the project without even having opened an issue that you missed these features. I think we could for example add a MariaDB module within mysql-haskell with just the mariadb specific stuff. Unless you had to make a lot of breaking changes then it could make sense.
The notion defined(𝑥) ⟺ 𝑥≠⊥ is problematic in my eyes because complex types have many elements between bottom and total elements. For example, 1 : 2 : undefined is a list distinguishable from ⊥. So defined = has WHNF? The definition I know from PCF (although I can’t provide a reference) is In flat primitive types such as Int or Int#, all elements except ⊥ are total. A function f :: a -> b is total if it maps total elements to total elements. total elements of (a,b) are pairs of total elements…
There’s no Haskell here, but I thought folks in this community might still find the topic interesting… Apologies if not / off topic - feel free to remove if inappropriate Dan's website – 8 Nov 26 What is static analysis, really? Many developers have come across terms such as “static analysis” and related ones such as “soundness” and “program semantics”, but what they mean and how they fit together isn’t always clear.

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 and trackpad scrolling as EventScroll (offsetX, offsetY) modifiers position instead of…
Does this give better experience then using the normal SQL drivers?
Done Start GHC 10.0 breakage inventory (anticipated) by andreabedini · Pull Request #148 · tomjaguarpaw/tilapia · GitHub
Haskell has lack of checking sub-types for sum-types in patterns. Termination checking is not a universal function - so it is hard to guarantee non-termination for any recursive function. Maybe, except the BUSY BEAVER’s number.
@winterland1989 is the author of GitHub - winterland1989/mysql-haskell: Pure haskell mysql driver · GitHub
I use the following. {\mathop{\mathrm{defined}}(x)} \iff {x \!\downarrow} \iff {x \neq \bot} \iff {x \mathbin{\, \mathop{\mathrm{seq}} \,} () = ()} \\ \mathop{\mathrm{total}}(f) \iff \forall x. \ {x \!\downarrow} \implies {f(x) \!\downarrow} Under this definition both id and absurd are total, while infinite data is defined.
This is great! I have been thinking about MySQL and MariaDB for some time now. beam used to have a MySQL backend, which I’d like to resurrect one day. MySQL (and by extension, MariaDB) have some interesting performance characteristics; for example, a connection to a MySQL cluster is much lighter than a connection to a Postgres cluster. Moreover, if you support the MySQL and/or the MariaDB wire format, then you can connect to something like Vitess. The other interesting thing about MySQL/MariaDB…