freenode/#lisp - IRC Chatlog
Search
12:30:09
pjb
aeth: on the contrary, functions and data structure must match. Otherwise you incur multiplied complexities!
12:31:05
pjb
aeth: for example, if you have a list, you must use a function that processes (first list) and (rest list). If you use a function that process the elements in random order with elt, then you get O(n) factors in your algorithm.
12:31:59
pjb
aeth: similarly if you have a vector, then you must use a function that processes (elt vector i). If you use (first-vector v) and (rest-vector v) = (subseq v 1), you get again a O(n) time and space factor in your algorithm complexity!
12:32:43
pjb
aeth: you may abstrat some uninteresting details of your data structure but the fundamental overall structure of it in your functions!
15:51:05
rpg
Has anyone tried to use Gary King's LIFT test library recently? I have fixed a bug in TRIVIAL-BACKTRACE, and was trying to run the tests, and now I'm going down a rabbit hole. Lift seems pretty completely bit-rotted. Tempting to just rewrite the minimal number of tests into 5AM...
16:24:45
rpg
scymtym: I looked at lift for a couple of hours, but fixing one bug just led me to another. My copy of TRIVIAL-BACKTRACE on gitlab.common-lisp.net (soon to be pushed) will use 5AM for testing. Tweaking the tests for compatibility was *much* easier.
16:50:36
rpg
Xach: Doesn't work at all with logical pathnames, and I get recurring errors where the standard config says "put test results in lift-results/" and LIFT says "I don't want a directory name here I want just a filename."
16:51:34
rpg
My guess is that LIFT was never tested with systems whose COMPONENT-PATHNAME was a logical pathname.
16:52:29
rpg
FiveAM is, AFAICT, not broken, so I just translated the tests for TRIVIAL-BACKTRACE. See https://gitlab.common-lisp.net/rgoldman/trivial-backtrace if you are interested.
16:53:08
rpg
(asdf:test-system "trivial-backtrace") now works, at least on SBCL. Haven't tested it on other implementations yet.
18:12:55
alandipert
is anyone aware of a JS impl. of CL with anything like DELIVER or SAVE-LISP-AND-DIE?
18:24:52
alandipert
parenscript and slip are 2 others i know of, slip is interesting because it includes VM and green threads http://lisperator.net/slip/. no longer being developed though, and impractical for me for a few reasons
18:29:19
alandipert
Xach i find most JS build toolchains are batch oriented and so antithetical to keeping as much tooling as possible resident in the target environment. so i'm curious about ways to develop lisp in the browser as "natively" as possible, including producing deliverables
18:30:09
alandipert
i am working on my own hobby lisp to this end, more of a clojure flavored lisp though. but i'd use a CL thing if one was available :-)
18:32:25
phoe
oni-on-ion: it's still very useful, just not a full Lisp implementation that you can run in the browser
18:43:50
aeth
pjb: What I mean is, if there's one part of the language where you're going to get near-C++ performance (well, probably matching a compiler that doesn't do SIMD optimizations) fairly easily, it's going to be arrays of numbers.
18:45:42
aeth
alandipert: JSCL is missing most of the fun parts of the language last time I looked, so I doubt it would have s-l-a-d.
18:47:10
skidd0
if anyone here has used cl-irc before, i'm curious if you've been able to verify that a user/nick is registered with NickServ
18:48:20
dlowe
skidd0: it depends on the server, but some of them have a mode where it prepends + to a nick when it's registered
18:49:52
skidd0
hmm, i'll have to check on that. I looked at find-user and tried to check the user's modes slot, but they were empty
18:57:22
dlowe
but if you explicitly auth/deauth and automatically deauth on disconnect, that handles most of your problems
18:58:03
skidd0
oh i see. so a nick "skidd0" joins the channel, sends a query over to the bot to auth, then can give commands
18:59:49
skidd0
well i'd be keeping a list of authorized nicks, so if they change nick, they lose auth