welltypedwit.ch
When the limestone of imperative programming has worn away, the granite of functional programming will be revealed underneath -Simon Peyton Jones
https://welltypedwit.ch
2,738 posts
2,240 followers
323 following
Getting Started
Active Commenter
comment in response to
post
yes!
comment in response to
post
yeah, how would you infer a type for
let f = \x -> x + 1 in f 5
without unification variables?
comment in response to
post
it's not "without unification" though if you're filling in unification variables ^^
comment in response to
post
if you want complete inference then yes, but even with an inference direction that's impossible without unification variables (because of lambdas)
comment in response to
post
i mean, that depends on what you mean by that ^^
if you restrict applications to applications on variables then no because you can just look up the function's type.
comment in response to
post
no, you can't recover the type of the function argument from the result. that's why in bidir application is the usual place where the direction switches from checking to inference
(assuming f is an arbitrary expression, not just a variable and you don't have unification variables)
comment in response to
post
i did not need to check who posted this ^^
comment in response to
post
i like commit. kind of like cut in prolog. it's also interestingly the opposite of (mega-)parsec where by default parsers only backtrack if they haven't consumed any input and you need to opt into full backtracking with try
comment in response to
post
jai will only have straightfunctors
comment in response to
post
i mean, there's nothing fundamentally wrong with using flying machines. maybe you just like walking over slime blocks any time you use your door like a psychopath
comment in response to
post
not sure how lsp-mode works but if you look under Output > Haskell in vscode, do you see an error like this?
also, are you using stack or cabal?
comment in response to
post
is it actually getting stuck or throwing "ABIs don't match" errors?
comment in response to
post
(that's my "niche tweet that 3 people will get" quota for the month)
comment in response to
post
in what way?
comment in response to
post
Haskell loves you too, little lesbian!
comment in response to
post
but that says absolutely nothing about *legal* concerns or literally any other reason to worry about copyright violations here
comment in response to
post
yeah. there is one grain of truth to it in that if you do believe that piracy is good, etc. (like many but definitely not all programmers), you can't treat this as a *moral* issue purely because it's an IP violation.
comment in response to
post
i think if you want to make an argument like that you'll need a little more evidence than vibes based on your personal experience
(and to be clear: i don't have more than vibes based on my personal experience either but i'm not the one making a sweeping claim here)
comment in response to
post
sure, an aGeNt will run a type checker on its own but that's like, such a low bar.
comment in response to
post
something that's always implicit in these kinds of posts is the assumption that generating, reviewing and correcting bad code written by an LLM is significantly more productive than manually writing it, without sacrificing the quality of the result
and i just find that pretty difficult to believe
comment in response to
post
i don't really have the patience to get into the contents too much right now but just the most important one:
comment in response to
post
i mean, you still can. you don't have to use google ^^
duckduckgo exists, i've been using ecosia ai-free for years (which usually has similar, sometimes even better results than google ime)
comment in response to
post
oh it was haskle! haskle.net
(there is also type-signature: type-signature.com)
comment in response to
post
i totally do but i also forgot what it's called ^^
comment in response to
post
that genuinely looks better but it also just looks like pastel netherlands lol
comment in response to
post
(not like that would have fixed this case because it was using sprintf but you get the point)
comment in response to
post
i mean, it's not as much the for loop that's an improvement as it's replacing the *quadratic* string concatenation ^^
(tbh this sort of thing kind of makes me think that js implementations got it right and strings should be ropes by default not byte arrays)
comment in response to
post
(i actually mostly meant the "slowing down everyone's builds in order to not solve the stability issue" part but yeah)
comment in response to
post
you can tell no one likes the bi flag because there’s no discourse about it
comment in response to
post
(it is really nice)
comment in response to
post
yes but it's also just kind of ugly tbh ^^
comment in response to
post
what did you want from megqparsec?
(also omg rainbow cat)
comment in response to
post
might work with fork_daemon? (if you make sure to handle cancelation correctly)
comment in response to
post
i kind of want to use flora for config more but it probably still needs a few more features for that ^^ (works pretty well for my blog though)
comment in response to
post
i mean, the haskell implementation isn't...
comment in response to
post
"there is a small chance that thousands of people will block you on sight but that probably won't happen, right?"
comment in response to
post
i'm sorry it uses what now? how does that even happen?
comment in response to
post
lmaoo
comment in response to
post
wow that account exists since january 7th (143 days ago) and they have posted 61K posts in that time. that is 426.5 posts a day lol
comment in response to
post
uses some cheap, low powered llm to generate a summary or something (that would explain the weird translations and quality) and then blasts them out at around one post a minute ^^ bsky.app/profile/dasg...
and the real goal is to get impressionable crypto idiots to click on the header images
comment in response to
post
omgg.
yeah i don't think there is a human involved in any of this ^^
pretty sure they have a server that continuously scrapes the web for articles vaguely related to crypto (with lots of false positives), ...
comment in response to
post
yeah, they also have one about ocaml and nix that is tagged ""bitcoin"" ^^
dasgeld.co/posts/3CFC04...
comment in response to
post
would also explain the cryptic urls