freenode/lisp - IRC Chatlog
Search
7:14:07
jasom
someone Fare? was asking about my ql2nix script. It needs a lot of cleanup, but here it is in all of its ugliness: https://github.com/jasom/ql2nix
7:16:49
jasom
Fare: oh, I want to hear more about safe killing of actors; that's generally not possible without a lot of restrictions
7:42:16
H4ns
fiddlerwoaroof: yason is not really trying to be very convenient for the use case that you describe, serializers for types. i've specifically wanted to be able to tightly control the json output rather than be able to serialize arbitrary lisp objects to json. this is not to say that your suggestion does not make sense, but there may be json libraries that are better for what you're trying to do.
7:58:41
schweers
properly implementing lisp numbers and the comparison of strings to numbers on javascript sounds like a real pain
8:02:14
pjb
schweers: there's no comparison of string to number in CL, so it should not be painful at all.
8:02:35
pjb
(there's only EQ, EQL and EQUAL which could apply both on a string and a number, and the result would be obvious).
8:03:30
schweers
often one cannot tell at compiletime which types the variables will hold, so what happens if I have (= a b) and one of them is the number 0 and the other the empty string? in CL this would signal a condition (I think). In JS it should yield true, right?
8:07:37
schweers
I would have thought that runtime checks are needed to implement this behaviour, yet JSLisp tries to avoid those
8:09:23
pjb
Anyways, if you want my advice, don't implement CL in javascript. Implement a CL to javascript compiler in CL. Then use CL to compile this compiler, then use this compiler to compile itself, then you will have a CL to javascript compiler in javascript.
8:09:48
pjb
This way, you only have to deal with javascript as a target language, not as an implementation language.
8:20:29
axion
sigh, JSCL devs still haven't fixed my bug reported where it's possible to freeze the entire image
8:28:22
edgar-rft
axion: javascript folks would be totally unhappy if there's nothing freezing your browser frequently
10:13:22
attila_lendvai
he's eye'ing infrastructure work ala tunes.org where he doesn't want to waste too much time on the hw related details of a compiler
10:13:48
attila_lendvai
btw, that red language mentioned in the comments seems interesting based on 5 minutes of reading: http://www.red-lang.org/p/about.html
11:19:40
Fare
some-use`, interchangeable? No, I believe Gerbil is superior for what I'm doing right now.
11:24:28
Fare
jasom, herep -- yes I was asking about ql2nix. As for killing actors, see my recent rant on the gambit list (if you follow me on twitter: http://twitter.com/ngnghm )
11:42:32
PuercoPop
slyrus: There is an actively developed and mantained fork of cxml: https://github.com/TBRSS/FXML/
11:59:31
some-user
Fare: Sorry, got disconnected. But what are you doing right know if it's not a secret? Or at what task Gerbil is superior to CL?
12:15:54
Fare
Gerbil has actors and RPC builtin, and a decent story for DSLs (including Racket-like #lang)
14:58:22
slyrus
Wow. There's a lot in that fork. I thought you meant a fork in the github sense, not a whole new project that uses cxml as it's starting point :)
15:56:11
jasom
minion: memo for Fare: it seems like all of your requirements for safe asynchronous aborts can be met by the CL condition system; a restart captures the total state at which the condition was signalled, such that higher levels of abstraction can install a handler that instructs the lower levels to continue until you are at a safe point for the higher level.