libera/#lisp - IRC Chatlog
Search
22:09:17
thymage
And I would like that error to be automatically resolved; I would like to clean away fasl files and completely rebuild the package.
22:13:25
thymage
But also... I'm getting compile errors on code when I'm using it as an asdf system that weren't there when I was using it in a fresh sbcl instance
22:33:05
thymage
So basically, I have a code fragment that works fine, a to z when I copy paste it in a fresh sbcl instance. But if I try to load my asd system using quicklisp with the softlink in place, I get an illegal function error
22:34:32
thymage
But I don't know why identical code would have an illegal function in one context as opposed to another
22:39:03
thymage
Like literally the strcmp function that it defines on the front page of the github repo.
22:45:04
mfiano
That's sort of expected. asd files are not compiled. and your lisp code has to defpackage with :use :cl and switch to your package with in-package
23:00:09
thymage
I'm happy with that. I only need to grammatically manipulate some of the language. I don't really want to change away from c-mera
23:03:05
thymage
Let's say I were going to make a minimal reproduction. Could I paste a couple of files, a lisp file and a system definition?
23:03:35
thymage
Oh ok, so I have to have both defsystem and defpackage. RIght now I only have defsystem
23:38:39
kagevf
thymage: I think you can add a condition handler to ignore errors ... and the condition system let's you add all kinds of features for condition handling and restarts
23:39:11
kagevf
you can think of the CL condition system like a superset of try/catch in other languages
0:21:47
jcowan
Blub in Lisp syntax is an idea that keeps appearing. I designed a few myself. Bottom Scheme is a tiny subset of Scheme that is fairly easy to compile to C, partly because it doesn't have Scheme's special magic: no call/cc, only restricted closures. CL/R is analogous, but for CL, and I have one for SQL scalar expressions. None of these have working implementations: ars longa, vita brevis.
0:25:17
jcowan
Then there is one to solve the problem of Scheme floating point: the default precision in almost all Schemes is double, so they can't be boxed without heavy compiler inference. Flopsy uses sigils for typing, and its types are floats and float vectors, procedures, and integers. The idea is that you write your code in Flopsy, which is valid Scheme, and debug it there. Then the Flopsy compiler converts it to C, and you access it through
2:22:44
jcowan
moon-child: Yes, but (a) nobody had thought of nanboxing then; (b) to this day, no Scheme implementation implements it