Search
Saturday, 27th of May 2017, 0:03:56 UTC
0:25:12
pookleblinky
aeth: There's a Tibetan language that has not changed its written medium through several vowel shifts over the last thousand years
0:26:14
pookleblinky
Compared to it, English has a reasonable spelling.
0:30:21
|3b|
english has lots of perfectly reasonable spellings... problem is that they are all mixed together :p
0:30:33
whoman
translations of tibetan texts are some of the most <adjective> english i have ever had the pleasure to read in this whole life
0:31:44
|3b|
(and probably like japanese, also has mixed in a few copies of same reasonable spelling from different time periods)
2:02:16
attila_lendvai
so, I've created one more zlib binding, but this one based on c2ffi: https://github.com/attila-lendvai/hu.dwim.zlib
2:03:06
attila_lendvai
with a random tester that seems to trigger zlib bugs when run in certain configurations... :/
2:05:31
attila_lendvai
ACTION finally goes to sleep... o/
3:02:40
beach
Good morning everyone!
3:03:52
minion
holycow, memo from jackdaniel: hey, could you drop me your email on query? thanks!
3:30:37
fiddlerwoaroof
I've been reading about the Nimble type inferencer for common lisp, does anyone know if the source code is available anywhere?
3:35:48
beach
But I think what we do in SICL/Cleavir is a better technique anyway.
3:36:18
fiddlerwoaroof
I should look at that
3:37:13
beach
We work on the HIR (High-level Intermediate Representation) of the program, and we describe type inference as a data-flow problem.
3:38:44
beach
And we use Kildall's algorithm to traverse the instruction graph.
3:39:46
fiddlerwoaroof
Are you familiar with the paper "Type Systems as Macros"?
3:40:33
fiddlerwoaroof
They're working in Racket, but they essentially show how you can use a macroexpander to check types and then erase them
3:40:35
fiddlerwoaroof
http://www.ccs.neu.edu/home/stchang/pubs/ckg-popl2017.pdf
3:45:10
fiddlerwoaroof
I'm still a complete novice to these topics, but I have a friend who's working on a lisp with Haskell's type system implemented as a racket macro
3:45:23
fiddlerwoaroof
s/macro/language/
3:45:49
beach
It does look like a nice technique for implementing typed embedded languages.
6:37:53
vtomole
How do i generate a random bit with the probability of getting a 0 being .40 and probability of getting a 1 being 0.60?
6:43:03
beach
vtomole: (if (< (random 1d0) 0.4d0) 1 0)
6:45:13
beach
Sure. But I reversed the probabilities.
6:45:39
beach
But you get the idea.
6:51:33
vtomole
beach: How is Cluster?
7:26:29
matthew__
** NICK Guest2279
8:08:25
beach
vtomole: No recent progress. I am too busy with my Earley-based lambda-list parser framework.
8:19:17
beach
Hey jackdaniel! Congratulations!
8:28:38
shka_
sooo, what /win means?
8:29:22
shka_
beach: i was reading SICL documentation
8:29:43
beach
What a surprise! Anything I can help you with?
8:30:02
shka_
i learned quite a lot about lisp from cleavier documentation
8:30:09
shka_
thank you for putting this together
8:30:59
beach
Sure. I am glad it was helpful to you. Anything in particular you found especially interesting?
8:32:00
shka_
well, i was looking for information about first class envs, but couldn't find it
8:32:21
beach
Ah. It is not in Cleavir, but in the SICL documentation.
8:33:18
beach
shka_: There is also a paper in Papers/Global-environments
8:34:06
paule32
why are binaries sooo bigggg? can i reduce the size by modular libs?
8:34:15
beach
shka_: Chapter 15 in the spec it looks like.
8:34:26
shka_
paule32: binaries of what?
8:34:51
paule32
clisp can produce bytecode, this can be converted to stub code of machine
8:35:28
paule32
a simple (print "hello") takes about 70 mega bytes
8:35:52
shka_
paule32: is it executable?
8:36:05
shka_
well, here is your answer
8:36:20
shka_
it includes lisp as well :-)
8:36:56
paule32
my old c programs have 32 kilo bytes - by using so - shared object file
8:36:57
joga
(shka_, /win is short for /window in irssi, ie. switches the active channel/query window)
8:37:16
shka_
it is sort of archive that not only includes your code, but also whole clisp
8:37:49
paule32
clisp feature is to use packages?
8:39:03
shka_
not sure what clisp does there exactly but from what i see, it does more or less what i said
8:39:45
paule32
i don't know other lispsers behavoir
8:39:59
shka_
perhaps it is the problem, maybe you can ignore this issue
8:40:20
shka_
there are multiple ways to think about it
8:40:51
shka_
i personally use cl server side, so i almost never care about binary size
8:41:19
_death
if you care so much about hello world applications perhaps C is the answer indeed
8:41:26
paule32
i have server here, which shall serve / host lisp projects, it has 24 giga byte ram with 2x 2 terra bytes hard disk space, ok, but the stream / makes the different
8:41:28
shka_
therefore, i have no idea how to reduce it
8:42:06
paule32
shka_: dynamic linking
8:42:24
paule32
there fore so/.dll are invented
8:42:49
paule32
to share/reduce memory
8:42:58
shka_
well, whatever, i'm not personally interested in that
8:44:05
paule32
ok, that should not be hard, but i had one eye on it
8:44:06
beach
paule32: It seems that no free Common Lisp implementation is organized so as to allow most of the code to be in a shared dynamically-linked library.
8:44:29
beach
paule32: I guess the developers have other priorities.
8:45:28
paule32
i don't understand the m$ they server software have muchhhh resource barbeque
8:46:10
paule32
on furtune, that my computers run linux
8:46:49
paule32
ok, back to the red line, this is off topic
9:10:06
gingerale-
** NICK gingerale
10:44:47
smokeink_
how to download all quicklisp projects to local drive ?
10:47:32
smokeink_
http://beta.quicklisp.org/archive/ <- access denied
10:47:55
smokeink_
can't view the listing with all files
11:04:33
smokeink_
hmm (ql:system-apropos "") seems to return all the systems
11:06:09
dschoepe
smokeink_: (ql:system-list) will give you that as a normal list instead of just to stdout. you could map ql:quickload over that, I suppose
Saturday, 27th of May 2017, 12:03:56 UTC