#lisp - Wed 14 Mar 2007 between 00:55 and 01:21

NY Lost Funds



pkhuonggigamonkey: #'(lambda ...) VS (lambda ...)
gigamonkeyAh lambda #'lambda
nyefgigamonkey: Yeah. Basically (lambda () ...) vs. #'(lambda () ...).
gigamonkeyYeah.
msinghnyef, thankyou for helping clear things up :)
nyefmsingh: No problem.
_3bRiastradh: where are you finding the char set in taocp?
nyefjaoswald: For the little things like object layout it shouldn't be too bad. For the big things about what spaces where, the details change over time.
jaoswald: The canonical reference for object layout in SBCL is the source, at src/compiler/generic/objdef.lisp.
Riastradh_3b, page 13something or so. I put it back in the bookshelf.
jaoswaldI had gathered that, but was not having enough fun with the detective work and was hoping it was explained in English somewhere with nice diagrams. I guess this is where I am supposed to volunteer.
_3bah, was hiding in a chart...looks like delta,sigma and pi in 3rd edition
nyefjaoswald: Now, a real trick would be to derive the diagrams from information available in the running system.
jaoswaldand create texinfo, PS, PDF, and HTML/PNG versions during the build process?
nyefAs part of the internals manual build process.
Admittedly, doing it that way would mean that the manual is specific to one particular port...
gigamonkeyAnyone know how to pass a pointer to a Lisp object through CFFI (or any FFI for that matter) so that you can get it back and convert it back to the Lisp object?
actionnyef could do it in SBCL, but that might not help you.
nyefcould do it in SBCL, but that might not help you.
baggitoyou could just convert integers to pointers and use an array or a hash map
gigamonkeynyef: so how would it work in SBCL?
Might help me find the right place in the CFFI docs ...
nyefsb-kernel:get-lisp-obj-address or whatever it's called, and there's an inverse in sb-di or sb-debug. Probably won't survive a GC if there isn't a reference on the stack, though.
gigamonkeyYeah, I was wondering about that.
ed-t8is there a lisp reference text which describes algorithmic analysis? (big O)
nyefOtherwise, if it's opaque to the foreign code, go with baggito's solution.
pkhuonged-t8: what has that got to do with lisp?
ed-t8pkhuong: nothing, but it would be nicer in Lisp
nyefed-t8: Why bother? Such analysis is fundamental CS junk, not anything for a reference book.
jaoswalded-t8: are you referring to a description of the behavior of the various Lisp operations?
ed-t8nyef: how else can i know how efficient an algorithm is?
Zhivagoed: big O just tells you the worst case
jaoswalded-t8: how is the algorithm described to you?
ed-t8Zhivago: so am I taking the wrong approach here?
pkhuongZhivago: an asymptotic upper bound, not necessarily the worst case.
ed-t8jaoswald: I don't understand the question
gigamonkeynyef: yeah. Seems stupid to have to keep an array of objects that have been passed through CFFI and clean it up and all that. Especially in the particular case I'm dealing with where I'm shoving some objects into an event object that passes through some C code and directly back out to some Lisp code on the other side.
jaoswalded-t8: the feeling is mutual.
jonathonnyef: you still having fun running sbcl on windows? ;-)
ed-t8jaoswald: lol, basically i'm trying to analyse how efficient the lisp code i'm writing is, and I thought(perhaps mistakedly) that analysing the algorithms i'm using is a good place to start
jaoswalded-t8: you know how efficient an algorithm is by looking at the cost of each step, and an analysis of the typical number of times a step is executed, and the typical cost seen at each step for the particular values encountered during the algorithm's execution.

Page: 4 11 18 25 32 39 46 

IrcArchive

NY Lost Funds