freenode/#lisp - IRC Chatlog
Search
6:57:31
beach
makomo: Thanks for questioning my decision of not wanting the compiler in the initial image.
6:58:26
beach
The "fast module" said (it is going to be too slow and take too much memory). The "slow module" (which is also lazy) believed the fast one.
6:59:12
beach
Yes, as I pointed out, if I can rely on the compiler being present, I can eliminate lots of code.
6:59:57
beach
I might have to be a little bit careful during bootstrapping. I don't want to save each byte of the X86 instructions in a 64-bit word. :)
7:02:05
beach
Anyway, thanks again. The current hypothesis is that I can include the compiler in my bootstrapping output.
7:02:09
makomo
beach: what is this effect you're referring to? i found at least 3 different ones described by Kahneman
7:02:47
beach
It's from the book "thinking, fast and slow". There are roughly two modules, the fast and the slow one.
7:03:19
beach
The fast one is lousy with arithmetic ans probability. The fast one can do that, but it is lazy and tends to believe the fast one.
7:04:29
beach
So, when we hear something like "if we fixed all dripping faucets in France, we could save 100 million liters of water every year", the fast module goes "oh, let's do that then, because that's a lot of water".
7:04:55
makomo
beach: awesome. interested to see how this will affect the bootstrapping. i have yet to understand the effective method computation thingies that are being discussed. i'm not familiar with that part of the mop yet
7:05:08
beach
The slow module, if it is not lazy that day, would divide by the population of France, resulting in 1.5 liters per person per year.
7:08:24
sixbitslacker
Here is your shopping list: (1) Copious dogboy annoyers; (2) Murderous bags; (3) Generous bum animals; (4) Satanic robber contributions; (5) Gaunt flanges; (6) Erroneous compressors
7:09:05
sixbitslacker
Here is your shopping list: (1) Dubious birds; (2) Satirical anti-shock mice; (3) Hesitant terrific selector units; (4) Anonymous dividers; (5) Chinese Unopened paradoxes; (6) Fantastic livers
7:10:03
sixbitslacker
Here is your shopping list: (1) Indolent exhibitions; (2) Recurrent wavechord skis; (3) Biochemical rigorous seducers; (4) Magisterial bears
7:11:13
no-defun-allowed
shinmera: thanks for lquery btw. very fancy clone of a js thing i'm glad i wasn't old enough to bother learning
7:14:38
no-defun-allowed
this had a nice big file of (useless group) spam and i used it to spam (useless group). very symmetrical.
7:15:11
no-defun-allowed
(useless group) is not relevant enough to be mentioned and is slightly controversial. you know how sbcl pprint says # if you go too deep? same thing.
7:15:24
MichaelRaskin
pjb: well, eval is the only standard way to evaluate macroexpand-1 in a lexical environment you can describe but cannot obtain as a lexenv object
7:17:02
MichaelRaskin
makomo: well, you might have missed that just having access to the list of names in the environment is enough to reconstruct a good enough approximation.
7:20:09
MichaelRaskin
makomo: agnostic-lizard does support expansion using an over-approimation of an environment now
8:10:39
_death
right, I'd expect a count instead.. then it's also easy to tweak the distribution with a "temperature" to create more/less surprising sentences