06:14 <dpk> diving more deeply into Racket for the first time, coming from the world of RnRS Scheme with some CL and a bit more Emacs Lisp, i'm curious: why does Racket's class system use a (send object method args ...) macro for ‘method’ invocation, ignoring the decades of Lisp work on single- and multi-dispatch systems which show it's essentially just another way of spelling (method object args ...) like any other Lisp procedure call?

06:16 <dpk> i know of only one disadvantage of the T/CLOS way, which is that both method and obj need to be defined in the namespace where you call the method, whereas with (send ...) you can theoretically have messages with any name without having to define them. but most CL'ers (and I) think this is a reasonable trade-off for a unified procedure call style which encompasses class method invocations, and in the case of CL for also getting multi dispatch

06:18 <dpk> (that disadvantage is reduced further by NO-APPLICABLE-METHOD, which is like Smalltalk's #doesNotUnderstand:, though it does still require that there's *something* defined with the actual name of the method)

06:22 <dpk> i also think it's particularly Schemey for things like public/private method visibility to be controlled by the regular module name export/visibility system, and with the Java style (for lack of a better term) that needs to be reimplemented at the object level

06:22 <dpk> anyway, didn't mean to turn this into a polemic, i'm curious why that design was chosen

07:30 <winny> I thought I read a paper discussing some of the choices behind racket's OOP implementation... I can't seem to find the one that introduces it

13:21 <random-jellyfish> is there a package with data collections like queues, stacks, maps, sets with efficient implementations kind of what you find in the C++ STL or the Java standard library?

13:22 <random-jellyfish> the mutable pairs/lists are so rudimentary

13:22 <random-jellyfish> they don't eve have methods to add/remove elements

13:22 <random-jellyfish> I mean I could write them myself, it's not hard

13:22 <random-jellyfish> but I'm wondering if there's something already there that I don't know about

13:24 <bremner> there's hash-tables, both mutable and immutable. IIRC they both have performance guarantees.

13:26 <bremner> vectors are the goto mutable container, although fixed length

13:28 <random-jellyfish> are there variable length vectors/lists?

13:29 <random-jellyfish> I could make my own using mpairs

13:29 <bremner> well, I mean lists are variable length, but deleting and inserting in the middle is not really idiomatic as far as I know

13:30 <random-jellyfish> I probably won't need delete/insert in the middle just at the ends

13:30 <bremner> well, racket lists are singly linked.

13:31 <bremner> (so inserting at one of the ends is expensive)

13:32 <random-jellyfish> if I use mpairs I would expect set-mcar! and set-mcdr! to be O(1)

13:32 <bremner> yes, that sounds right

13:33 <bremner> there's pfds, but that's more about immutability

13:33 <bremner> https://docs.racket-lang.org/functional-data-structures/index.html

13:41 <bremner> oh right, and https://docs.racket-lang.org/data/index.html

13:43 <random-jellyfish> ah yes, gvectors, there they are

14:49 <random-jellyfish> in DrRacket is there a keyboard shortcut to delete everything between a pair of parentheses?