freenode/#lisp - IRC Chatlog
Search
2:41:07
pjb
(let ((a '())) (pushnew 'test (aget a :foo)) (pushnew 'test (aget a :foo)) (pushnew 'test2 (aget a :foo)) (aget a :foo)) #| --> (test2 test) |#
2:41:32
krwq
thanks pjb! I thought there was something inside cl but assoc didn't work for me - this looks good though :) thank you!
2:46:29
Timzi
when I'm using the (time form) macro to get a runtime of something, how do I capture that info? Describe says it's printed out to *TRACE-OUTPUT*, how do I get to that?
2:49:00
pierpa
(let ((*trace-output* (make-string-output-stream))) (time (+ 1 1)) (get-output-stream-string *trace-output*))
10:35:45
ebzzry
Aside from the computer language benchmarks game, are there are benchmarks between SBCL and GHC?
10:49:18
jackdaniel
I didn't know GHC can compile Lisp code (because I'm sure SBCL can't compile Haskell code :-)
10:51:37
beach
I logged in to EasyChair today, and it seems ELS2018 has not yet been registered, or am I missing something?
10:54:15
p_l
would a drakma replacement that depended on C library be welcome, if the dependency was designed to be factored out later without changes to user applications?
11:02:15
Shinmera
There's two issues I'm aware of: 1) it's slow 2) it doesn't properly handle unicode filenames in multipart requests
11:04:01
Shinmera
Cthulhux: It would need a significant rewrite to the point of practically being a replacement.
11:05:25
Cthulhux
given that HTTP moves rather slowly, there is probably no need for a hyperactive development around that either
11:06:19
p_l
Cthulhux: Yes, HTTP/2 *recently* solidified. And requires *significant* redo because the whole model changed significantly
11:07:39
beach
p_l: Are you saying that a rewrite that depends on a C library would require significantly less work than a rewrite that doesn't?
11:08:37
beach
But it would be possible to design the protocols such that the C library could eventually be replaced?
11:08:59
p_l
I have a bunch of "short coding things" that I'd love to do in CL, but they usually end up differently because there's a lack of necessary libs which are *not* short undertakings, even for the bare minimum of features
11:09:27
p_l
even that it would be easy for Joe Random Hacker later on to swap implementation when they need some custom tweaks
11:13:22
Shinmera
One thing that's still somewhat higher on my todo is bindings for bearssl so that we can finally have an easily shippable SSL implementation that doesn't depend on openssl
11:14:49
Shinmera
I don't know too much about libressl, but bearssl has zero dependencies and no malloc/free calls, and other things about it that are very promising.
11:15:14
p_l
Shinmera: in scope of similar ideas was to replace current cl+ssl with pluggable implementations
11:15:52
p_l
libressl is possibly cleaned up but with arseholes in charge who don't really care about devs outside of OpenBSD
11:16:09
Shinmera
Particularly the no deps thing is very interesting to me since that means I can just ship it easily myself.
11:16:48
p_l
Shinmera: it could be possibly bundled into one C file that could be compiled by ASDF in presence of compiler and easily precompiled for others (like SQLite is)
11:17:57
p_l
ideally, I'd like to have some common libs (some ediware) replaced with options that are "pluggable" in terms of implementation
11:18:47
Shinmera
I'd be fine if the pluggability was simply implemented by offering alternative packages that offer the same interface otherwise, so all you had to do was either replace the :use, or s/r whatever:
11:19:25
p_l
Shinmera: I was thinking that, + maybe a compat package that provides "classic" package names
14:10:11
phoe
cuso4: you might want to ask around #lispgames - they're much closer to the GPU than #lisp