freenode/#sicl - IRC Chatlog
Search
9:22:30
no-defun-allowed
what's weirder about that is that it hasn't been on wifi for the last half hour.
12:12:24
heisig
beach: I just sent you a pull request that fixes some (minor) issues in the SICL README.
14:39:56
makomo
beach: wow, that paper looks very exciting. once again (as i've said before), i like your mathematical style (all of the stuff neatly formalized into definitions, etc.) :-)
14:43:31
makomo
that's alright, i haven't read the thing yet anyway. and right when i decide to get back into the CL loop, midterms hit me (the next 2 weeks) :-)
14:46:44
makomo
heh, it had to happen sooner or later, but i always find it interesting how i get interested in something right before certain deadlines (either lab exercises or exams). a lot of other people have told me the same thing happens to them as well
14:54:56
beach
I think it is psychological. When I was at the university, I always got this irresistible urge to clean my room when I had to study for exams.
15:04:45
beach
No I didn't. In fact I don't think I am that fast. But I use a lot of tricks, like abbrevs (normal and dynamic).
15:05:22
beach
I do have the habit of writing papers, specifications, documentation, etc., so the flow is pretty good.
15:07:56
beach
Like, here on IRC, I have short abbrevs for things like "Good morning everyone!", "Common Lisp", "Common Lisp HyperSpec", "first-class global environments", "(first) Climacs", "Second Climacs", "(admittedly small) family", "/msg nickserv identify", "/msg nickserv ghost beach", etc.
15:10:14
beach
Also, my fingers seem to be unable to type the words "the" and "language", and instead "teh", and "langauge" show up, so I define the latter to be abbrevs for the former. Saves a lot of backspace.
16:02:26
scymtym
beach: i would like to make ECLECTOR.READER:WRAP-IN-QUOTE (analogous to ECLECTOR.READER:WRAP-IN-* for the quasiquotation constructs). i discovered the need while working on the client i showed yesterday - it wants to use its own symbol token representation for everything. does that sound good?
16:41:23
scymtym
an alternative design would be to have the QUOTE reader macro call INTERPRET-SYMBOL, but that would be less flexible (as in subsumed by the current suggestion)
16:42:43
scymtym
ok. it worked for my use-case and the README still warns about breaking changes so we can go back or change it later if we must
16:44:09
scymtym
the other changes (see #clasp) may be more problematic but as they wouldn't hurt clasp, i would still like to go through with them now and make the protocol consistent
16:54:35
jcowan
beach: if I ever get around to writing a "C with classes" language, I will probably call the compiler Second CFront.
17:02:03
jcowan
Only if you believe in historical inevitability, or believe that C itself is a terrible idea.
17:03:35
jcowan
btw, beach, if I were you I'd implement short-floats as BF16 and the devil with the silly requirements of the CLHS in this area (it requires 19 bits, which never made any sense).
17:35:16
jcowan
Well, of course you must do what you please. At first I thought the short-float restrictions were inherited from the PDP-10 with 18 bits, but I was forgetting about the sign.
17:36:03
beach
I was thinking of only implementing single and double, where single floats are immediates.
17:40:52
jcowan
I suppose it depends on the purposes of SICL (not your purpose in writing it, but the purposes which the users who choose it will have(.
17:41:46
beach
My goal is for it to be an excellent implementation for development, and then I want it to be the basis for CLOSOS.
17:42:15
beach
But I definitely have plans to make it as fast as possible without sacrificing safety.
17:43:19
beach
Improved safety and security compared to existing OSes, but also a superior multi-user development environment. Like Genera, but safer and multi-user.
18:49:27
jcowan
(that is, single target language; compilers for other languages are of course possible)
19:21:06
jcowan
beach: I now note that the size limits for floats are only recommendations, not requirements