freenode/#lisp - IRC Chatlog
Search
4:22:53
bqv
Context: would like to port this https://github.com/NixOS/nixpkgs/blob/8585991bfb629edda1e42c191bef935d9d70d690/pkgs/development/lisp-modules/clwrapper/cl-wrapper.sh#L107
4:27:15
fiddlerwoaroof
It looks like, if CFFI is loaded, that might just work? https://github.com/NixOS/nixpkgs/blob/8585991bfb629edda1e42c191bef935d9d70d690/pkgs/development/lisp-modules/clwrapper/cl-wrapper.sh#L102-L103
4:28:11
bqv
Ah, hm. Maybe it's just not pulling through then... can I safely remove the #+cffi to test?
4:31:19
fiddlerwoaroof
https://github.com/NixOS/nixpkgs/blob/8585991bfb629edda1e42c191bef935d9d70d690/pkgs/development/lisp-modules/clwrapper/build-with-lisp.sh
4:31:43
fiddlerwoaroof
Looking here, it looks like they support sbcl/ecl/clisp, maybe just add a case here for ccl?
4:33:30
bqv
What would the entry there contain? ccl is already supported, per my vague understanding, and I get packages to build
4:34:42
fiddlerwoaroof
Yeah, I think that's right, this looks like ccl: https://github.com/NixOS/nixpkgs/blob/8585991bfb629edda1e42c191bef935d9d70d690/pkgs/development/lisp-modules/clwrapper/setup-hook.sh#L18-L19
4:35:51
fiddlerwoaroof
I think you might try making sure that cffi is always loaded, and that code might cleanup the shared libraries
4:37:19
fiddlerwoaroof
Otherwise, look here: https://ccl.clozure.com/docs/ccl.html#the-foreign-function-interface
5:44:25
bqv
Still no dice. Not sure what's wrong. Don't quite understand how this works well enough to debug
8:22:56
phoe
my schedule is kinda f#<...>d up after these few days and I still need to record the talk
8:34:20
beach
I am sure you will. Sorry to hear about your complicated schedule. I hope you recover soon.
9:00:45
ck_
phoe: so 13:00 CET? I thought I'd read 12 before. Anyway, can these be put onto the meetup schedule on planet.lisp in the future?
9:03:07
phoe
ck_: https://mailman.common-lisp.net/pipermail/online-lisp-meets/2020-November/000046.html
10:33:20
easye
phoe: talk is how long? (trying to time cooking and family Zoom stuff between the Talk and the Hangout)
10:42:21
mfiano
Hmm, is there a way to #'fill an adjustable vector which has a fill pointer less that array-total-size with NIL's efficiently (which a compiler could just memcpy or something)?
10:45:46
mfiano
On SBCL it honors the fill pointer. My use-case is to free references to objects beyond the fill pointer
10:47:03
mfiano
Yeah that's what I'm doing. Just wanted to be sure there wasn't a better way (this is in a very tight loop)
10:48:19
beach
mfiano: In a tight loop, you still have to assign to every element, so unless you have very few elements, moving the fill pointer is not going to make a difference.
11:00:21
_death
you could keep a displaced array and fill that, though I'm not sure that would be fast
11:02:55
mfiano
Most compilers struggle to optimize displaced array access in my experience. I didn't dig too deeply though
11:05:40
phoe
my heap implementation in damn fast priority queue emulates a fill pointer via an ub32 + simple array
11:15:05
jackdaniel
and given that you use ub32, that could not work (because on 64 bit implementation it will probably have 61 bits)
11:29:58
no-defun-allowed
The CL numeric tower needs to support numbers written in up-arrow notation, so I can write (make-array (↑↑↑↑ 3 3))
11:30:04
aeth
most-positive-bignum is old news. My favorite constants are integer encodings of various mathematical proofs. In particular, the Riemann hypothesis, the P vs NP problem, or any other Millennium Prize problem. If anyone can find these constants, please share here first.
11:30:44
aeth
Good thing we have bignums. I doubt any of these proofs could be encoded in a way that fits in a fixnum unless you wrote a very specific compression algorithm.
11:36:35
aeth
To be fair, I did not specify a restriction to unsolved Millennium Prize problems and one has been solved, so in theory, the (uncompressed) bignum exists.
13:38:26
phoe
https://www.youtube.com/watch?v=V4P9lFK79hQ Control Flow in Common Lisp - Online Lisp Meeting #11, 26.11.2020
13:41:40
Josh_2
I've had the book a few weeks now, that talk convinced me to take some time to read it
13:47:18
phoe
ghard: I have, and I enjoyed it. I attempted writing a swank server in it once but didn't really finish it since it wasn't important enough to me at the time.
13:48:34
aeth
no-defun-allowed: it's the kind of thing that's hard to make portable, just because the portability libraries are probably insufficient (you just have bordeaux-threads, really)
13:48:41
no-defun-allowed
I also have a thread supervision library. If you give each job a mailbox, you would have actors.
13:49:42
aeth
no-defun-allowed: it's the thread part that might be insufficiently portable, if you need to get really low level
13:50:13
aeth
no-defun-allowed: but maybe it's feasible; you probably know more than I do about that part of the ecosystem