libera/#sicl - IRC Chatlog

12:14:44 scymtym relaxed clause ordering and more flexible accumulation would be very useful in my opinion. collecting into vectors or other data structures could be one feature. collecting in arbitrary sub-forms (like iterate allows) could be another. maybe via loop-collect like we already have loop-finish

12:39:31 yitzi beach: sure, I'll take a look at Salmagundi

12:40:30 yitzi scymtym: Khazern already has some relaxed clause ordering. Termination clauses like REPEAT, WHILE, etc can be intermixed with FOR/AS and WITH.

12:41:03 yitzi The collection ideas are good. I'll think about them.

12:42:34 yitzi Also, I was thinking of adding a CLEANUP clause that works like FINALLY but puts the forms into UNWIND-PROTECT. FOR/AS clauses can do that vie KHAZERN:WRAP-FORMS, but the user cannot. For example they cannot do cleanup on WITH bindings.

12:43:32 scymtym yitzi: that's good. still not sure what can be done pratically since conforming code can't rely on any of the extensions. maybe the situation will be like that of package-local nicknames, so one or more semi-standard extensions, at some point

12:44:36 yitzi For the iteration extensions, I've added features like :loop/iteration-path/elements

12:45:52 yitzi And khazern-extrinsic can be used as the primary LOOP in a single system via SHADOWING-IMPORT-FROM.

12:45:53 scymtym a similar issue i have been sad about for years is format's ~{...~} not accepting sequences but that would be for

12:46:05 scymtym invistra

12:46:29 yitzi That would not be hard to add to invistra.

12:46:56 scymtym oh yeah, extrinsically using sicl modules could be a pratical way

12:47:36 yitzi I added a section on it recently https://github.com/s-expressionists/Khazern/?tab=readme-ov-file#replacing-builtin-loop-with-khazern

12:48:53 yitzi If you imprort khazern-extension-extrinsic instead of khazern-extrinsic you'll get the extension stuff.

12:49:29 yitzi Err, actually I should probably add details for that also, since it may not be obvious

12:50:01 yitzi Since the import doesn't change but the asdf dependency does.

12:51:41 scymtym (cl:defpackage #:the-real-cl21 (:use #:common-lisp) (:shadowing-import-from #:khazern-extensions-extrinsic #:loop #:loop-finish))

12:51:46 scymtym i'm sorry

12:52:39 yitzi No, sorry for not being clear. It is still import from khazern-extrinsic, but the asdf depends changes. I'll come up with an example

12:53:04 scymtym i was joking

12:53:52 yitzi oh, I get it now. the package name. Nice

13:01:01 yitzi Updated that section https://github.com/s-expressionists/Khazern?tab=readme-ov-file#replacing-builtin-loop-with-khazern

13:03:07 yitzi duuqnd: I'll think about that one. You could probably do with for-as-arithmetic followed by for-as-equals-then using MOD. For the 2d array, I suppose it would be useful if there was an inverse array-row-major-index

13:14:32 duuqnd yitzi: Just so I'm being clear, the loop I would want to be able to produce would be roughly equivalent to (dotimes (y 240) (dotimes (x 320) [...])). I guess I could do that by doing modulos on a total index counter, or making y's :THEN be (+ (if (zerop x) 1 0) y), where x is then getting modulo'd before y gets updated... but that doesn't feel as clean... but maybe I don't actually need what I think I do.

13:17:28 yitzi Yeah, its messy

13:18:31 yitzi Are you storing it in an array?

13:24:16 duuqnd typically, though the important part is having the correct x and y

13:24:51 duuqnd sometimes the array may be 2D, sometimes it's 1D and it's just intended to be used as an image

13:26:13 duuqnd Or sometimes I may want to walk a 2D grid and :COLLECT things from it. I can imagine quite a few uses, though nesting LOOP does work so I can't complain all too much. It's just clunky.

13:29:16 yitzi Yeah.

13:29:33 yitzi For the array specific stuff: https://plaster.tymoon.eu/v/7BWFOYEYY#4884

15:52:29 ingeniot1 ** NICK ingeniot

18:17:27 yitzi beach: I'll start working on Salmagundi on a separate branch so I don't break SICL. It will take me a while to figure everything out, but I suspect that system reorganization will help avoid a lot of the duplicated intrinsic/extrinsic system declarations and lisp code.