@Bubbler you can pass a function to another "function" (though it's called an operator) - but you can't pass an operator to an operator (i mean of the same kind, e.g. mop to a mop)

so, that's just pushing the problem to the next level in the function/mop/dop hierarchy

@ngn A mop has different type from a function, so it's not a problem. But the true problem is we can't pass an op to anything (which gives rise of hyperators)

@Bubbler well, if hyperators could take other hyperators as hyperands.. :)

they would be the thing we call functions in functional programming

wouldn't it be simpler to allow functions as first-class objects, instead of going through this circular reasoning?

or stop advertising apl as a functional programming language

@Adám typo in dfns.notes.hyperators: "worspace"

Hmm.

We should (re-)think about the syntax then

And even if we admit that APL isn't a true functional language like Haskell, we can claim that it supports functional paradigm, like Python and JS do.

@Bubbler sure, but that doesn't mean much

js is actually a scheme in disguise :) the story goes that Brendan Eich initially wanted to put scheme in netscape but management specifically requested something with a c-like syntax, so it looks familiar to c/java programmers

I imagine, if we get away from the current "tacit" programs and the parsing rule dedicated to operators, and mark every (possibly higher-level) function as either monadic or dyadic, we could have a functional language that supports basic APL syntax.

Though we'd need to experiment how many parens we'll need

@Bubbler designing the syntax is the easy part. persuading dyalog is harder.

@ngn I don't believe Dyalog will accept it anyway, and I don't think it can be called APL either. At best it'll be an APL derivative.

i don't think apl+namespaces should be called apl either..

some people wouldn't even consider dfns to be apl

@ngn I need to move dfns to GitHub…

@Adám isn't it a ws (i.e. not source code)?

@ngn It is a ws. So?

@Adám well, you know, it's a binary blob, can't be diffed.. there's not much use putting it under version control

@ngn I'm converting it to one text file per item, and then "building" the ws from those files.

@Adám in that case it makes sense. the ws is just a build artefact.

Yup. This is already the case with the conga ws.

cool