freenode/#lisp - IRC Chatlog
Search
8:33:37
flip214
hmmm, so I need to provide a project "common lisp open source software using recommended evals" or something like that - and call it cloßure, I guess.
10:12:21
Cymew
lukego: About that discussion yesterday about quicklisp, asdf and finding files to load, have you seen this: https://common-lisp.net/project/asdf/asdf.html#How-can-I-debug-problems-finding-ASDF-systems it might be of some help, hopefully.
12:47:06
easye
Cymew: thanks for the docs on ASDF. Definitely a better starting place than existed before for users to get useful information.
14:13:09
pfdietz
I could use a javascript in CL. There's an implemenetation described in cliki, but it's like nine years old.
14:18:18
p_l
parenscript is about easy writing of JS from CL. There's also JSCL which attempts to implement a proper CL in JS
14:36:44
Cymew
easye: I need that myself sometimes, and fare was kind enought to provide some debug suggestions which Robert has now put in the manual. That is good, as I will forget my note about it... ;)
17:36:42
anlsh
Hi, can anyone tell me how exactly I can submit a patch to the Alexandria gitlab? I went over to the merge requests tab, saw the "email new merge request" tab, clicked it, and it gave me an email.
17:37:30
anlsh
It seems like I need to send an email to the generated address with a subject of the url to my branch
17:38:14
anlsh
But also the "fork" button on the project is disabled, so do I just need to create a bare repo to push it up?
17:38:35
_phoe
anish: the fork button is disabled if you have hit the max project limit on your account
17:45:04
anlsh
alexandria's readme should also be updated too, since it still says to submit patches to the mailing list
17:56:07
anlsh
More of a lisp question in general, are you not supposed to export new symbols when updating a package?
17:57:15
refusenick
The "cells" library has me thinking: would it make sense to have a graph database consisting of queries to a network of CLOS objects?
17:57:29
refusenick
I suppose it's more interesting as a thought experiment than something practical.
17:59:50
Xach
Hmm, I'll have to dig a little. I can't find it from a quick googling. It's on my bookshelf somewhere.
18:00:21
Xach
Joe Marshall also wrote a series of blog posts about versioned persistent CLOS objects in a project called ChangeSafe. I think the code (not finished?) is public for that, too.
18:00:41
refusenick
Lisp is a compiled language, but for the kind of long-running, distributed systems becoming more and more common, a homoiconic representation seems ideal for runtime optimization. Is there any work on CLOS and partial evaluation?
18:02:45
refusenick
Xach. Yeah. Automatically tracking versioned objects and picking up on optimizations missed by the programmer could be a good way to compete with static languages.
18:03:50
refusenick
I know I'll get flak for this, but I can't see CL being the language of the future because it doesn't have security.
18:04:26
refusenick
I know there's ongoing work to support it via first class environments, but it might be easier from the ground up, with support for CL built atop.
18:04:58
_phoe
this is a big sore point, correct, and first-class global environments are meant to solve exactly this.
18:06:43
refusenick
Might it be easier to build a language wholely out of first class environments a la Kernel or Io and provide a compiler to this from a minimal core of CL (tagbodies and other primitives) to compile each module into its own proper sandbox and fix interop issues as they arise?
18:07:35
_phoe
Or rather, you don't need to build a language wholely out of FCE - you simply need a way to chroot like in all unices.
18:09:38
Xach
I am also not especially persuaded by "if you don't do X lisp will die", because Lisp has been dead for a while.
18:09:47
Shinmera
This is probably gonna get drowned out in the current debate, but I made some good progress with Alloy and am really starting to like the widget system I came up with. https://www.youtube.com/watch?v=8-NJrJqDkrQ
18:12:00
Shinmera
It's a chain of Alloy -> Simple -> SimpleGL -> Trial (my game engine). You can also use an independent GLFW backend for the last part, or implement the Simple interface for different renderers.
18:13:24
Shinmera
The GLFW backend allows making multiple top-level windows, but Trial does not (because I don't need it there)
18:13:42
refusenick
Shinmera: I was about to dive into learning everything about CEPL to make an animation backend for Maxima, but now I'm conflicted!
18:15:44
Shinmera
The SimpleGL extension it provides and I use directly uses GL though, whereas CEPL wraps all GL access in a lispy framework
18:16:03
refusenick
CEPL is probably what I'm looking for, then. Maxima is perfect for me until I want to animate something.
18:16:46
refusenick
(kind of a problem since I want to study dynamical systems, plus I'm taking a class on differential equations next semester)
18:18:30
refusenick
A lot of Lisp projects seem to work out of the box with SLIME. I use Sly, though, because I like the defaults. I remember having some issues connecting to Stumpwm when I used that. Anyone here live-program Maxima with SLIME or Sly?
18:23:09
refusenick
I might use Alloy if the Haskell animation I'm working on doesn't pan out. I briefly tried to animate things in a Jupyter notebook with Python as a backup, but it didn't work. Maybe I've just gotten used to Emacs, but Lisp works correctly on first tries than most other languages. I didn't go for it this time because I wasn't aware of a ready-made high level visualization library!
18:36:55
refusenick
If a Lisp program restricted itself to CLOS objects (no non-object primitives like car and cdr - not everything is CLOS, right?), could it be 1-to-1 mapped to a pure OO language like Dylan?
18:38:41
refusenick
A more relevant question, I suppose, is whether CLOS and related systems have a equational theory.
18:42:25
refusenick
Shinmera: For what it's worth, unrestricted message passing also appears to be in a rut equationally - Io's semantics are equivalent to Kernel's, IIRC, and we still don't know how to AOT compile that (mostly because there's no incentive to study it)
18:43:00
refusenick
More restricted message passing systems like the pi calculus have equational theories and can be compiled though, IIRC
18:43:36
refusenick
Are the capabilities of reflective multiple dispatch OO systems a strict superset of message passing objects?
18:47:59
refusenick
I remember reading that Shutt proved that unrestricted reflection a la 3 Lisp (the real target of Wand's paper against fexprs) truly does have a trivial equational theory because the ability to jump between levels exceeds what continuations can represent, or something like that.
18:48:35
refusenick
Clearly, CLOS is compiled, though, and I even remember seeing papers on optimizing it.