06:10 <discocaml> <leostera> anyone's got a snippet around for applicative functors for building up a function type BUT written in a way that you can manually step through the chaining?

06:10 <discocaml> <leostera> i figured i'd ask before bashing my head against this for a bit but my gut tells me i'll want to model it as composition with a gadt

06:16 <discocaml> <leostera> i got the basic `app : ('a -> 'b) t -> 'a -> 'b t` but building this by means of functions alone means i end up with a single opaque function that i can't manually step

06:17 <discocaml> <leostera> and i want to be able to decide _how_ the stepping will happen at a later stage

06:44 <discocaml> <functionalprogramming> what does manually stepping mean

08:07 <discocaml> <octachron> The GADT option would be simpler to prototype, but you could also have a function `step: 'b t -> bool` that does one step

09:13 <discocaml> <wokalski> Is there a way to detect if a server closed a TCP connection with eio?

12:00 <discocaml> <leostera> thanks @octachron

12:01 <discocaml> <leostera> i’ll try prototyping the gadt but something tells me this is gonna be awkward

12:02 <discocaml> <leostera> in short every step is used to direct a deserializer and produce a value that will be passed as an argument to the final function

12:03 <discocaml> <leostera> the last step then pops off the value that you were constructing (so it finishes the call chain)

12:22 <discocaml> <octachron> This sounds very straightforward to do with a GADT. But with your description doesn't require control over steps.

14:47 <discocaml> <leostera> that control appears when the the deserializer format needs to make decisions based on how many things to deserialize, or what to do between elements

21:59 <discocaml> <barconstruction> I use a lot of Python at work and recently I started searching to see if people had used OCaml successfully in an ergonomic way for scripting tasks. I found a blog post by Yaron Minsky basically saying this is feasible if you compile custom toplevels with a bunch of libraries baked into them so that they're available immediately. Then I stumbled across a post in the C subreddit where somebody was like "I can use C as a scripting langua

22:00 <discocaml> <barconstruction> Felt like this C programmer was sent to remind me that you cannot trust programmers to be neutral about this stuff haha

22:01 <mi6x3m> hey, can anyone tell me if it's normal that OCaml 5.0+ produces no .a library for bytecode only ocamlmklib

22:01 <mi6x3m> I am building a project and when I use ocamlmklib with .cmo and .o files it produces a .cma but no static library any more

22:01 <mi6x3m> from what I read it should give both?

22:02 <discocaml> <barconstruction> https://www.reddit.com/r/ProgrammingLanguages/comments/1ahmkq0/using_c_and_other_languages_as_a_scripting/

22:02 <discocaml> <barconstruction> > it gives the same "feel" as if using a language like Python, Perl, Awk, etc.

23:02 <discocaml> <functionalprogramming> lmao

23:22 <discocaml> <druimalban> They've presumably never heard of csh