| |
| |
| |
|
Page: 1 2 3 4
Comments:
<0> Hmm. I really don't see what the advantage of keeping stuff in allocation-order is.
<1> it improves locality
<0> Does it? I thought re-ordering stuff by trace-order improved locality more.
<1> rydis: I don't think that's true. According to Paul Wilson, the best predictor of locality is the order in which a program allocates objects.
<0> I see. Thanks!
<1> In the GC algorithm that I am thinking about off and on, I also want to preserve the order of allocation in the nursery, but that is for a different reason, namely it allows me to have a much better idea of the age of an object, and so it gives me a more precise way of promoting objects.
<2> good night
<3> evening
<1> hello Xophe
<4> hello... maybe it's OT question... does someone installed the FixedMediumLispm-13.pcf.gz font, posted in these days on c.l.l., on Mac OS X and Emacs, compiled with Carbon option on (and X option off)?
<5> I'm not coming up with any good ways to do a precise scav of the x86 register set. :-/
<6> should be easy, just emit register location->variable mapping as part of the compilation process
<6> you need that for a good debugger anyways
<7> foom: do you?
<6> well, don't you want to be able to inspect things that happen to be stored in registers?
<6> I know I sure do
<7> there are multiple optimisation and debug settings
<7> and you can access the arguments to a call without knowing the layout throughout its execution
<6> sure, you can write a crappy debugger without needing that.
<8> hmm
<8> are alot of you people working on the machines themselves rather than just using them ?
<1> jonaslund: many of the participants of this channel work on implementations, if that's what you mean.
<8> sounds fun
<8> i was in here a while ago, never had the time to ask one thing i'm kind of interested in
<8> linear logic? opinions
<8> (as in the baker papers)
<7> try #concatenative
<9> pkhuong-: not related
<5> foom: The problem is that when you get down to the level of VOPs, there's no precise mapping of when in the instruction stream a value goes live, and there's also the possibility of the same TN being used for both a boxed and an unboxed value.
<7> kpreid: in the baker sense? Pretty.
<9> not really. most concatenative systems have "dup"
<7> Which explicits non-linearity.
<9> hm, but my initial statement was excessive
<9> you're right
<6> nyef: I'm not saying it wouldn't be hard to implement, just that it's "easy", and would be desirable to have for other reasons. :)
<1> nyef: are you saying that SBCL by design cannot determine what registers contain live data and which live data is boxed?
<5> beach: Not on x86.
<1> sounds like the design needs to be modified
<6> on other platforms it knows because the registers get segmented.
<6> which isn't really ideal either.
<5> Hrm...
<5> I'll have to look at the problem further.
<5> Just being able to get a precise stack scav is an improvement.
<10> hmm... too bad there isn't a clisp shell repo in ubuntu
<10> s/repo/package
<11> xarq: are you sure? May be in Multiverse
<11> there is at least a sbcl package
<10> I mean clash, not clisp itself
<11> I have a design question. I have some byte-data which I convert to a list of cons. The function looks something like (loop do (let ((length (blackbox0 byte-data)) (append list-of-cons (blackbox1 byte-data pos length)) (incf pos length))). But actually the only thing I do with the list of cons is looping through the list. Checking Something and Create a new list of cons. Which I convert in another function that loops through the list of new cons
<11> into byte-data
<11> I don't like it that I'm using 3 loops where I could actually use 1 loop
<11> So would it be the best way to give the first function a function parameter which it can execute after converting a cons
<11> or what would be the best way in lisp to do it?
<1> I'm afraid I don't follow your description of what it is you want to do, or what it is that you did.
<12> how do you print a tree structure nicely formatted without parens?
<1> kingruedi: for instance, your APPEND has no effect that I can see.
<1> kingruedi: also, it is inefficient to add to the end of a list. It is much better to add to the beginning, and then reverse the list.
<12> is there a format directive to do this?
<1> kingruedi: and if you construct a list by successive calls to some function f, and then want to apply a second function g to all the elements of the resulting list, then you might as well do (g (f ...)) in the first place.
<1> marcelino: I don't think so
<12> i am ***uming that to print a list tree structure nice involves uses car/cdr combinations to traverse the tree?
<1> sounds very likely, yes, depending on how your tree structure is represented, of course.
<13> kingruedi pasted "3 Loops => 1 Loop" at http://paste.lisp.org/display/23675
<11> beach: please see the paste. It should explain my problem a little bit better.
<1> ok, hold on...
<11> (I know the first function is buggy. Please ignore that. The real function looks different)
<11> But my problem is. I have 3 loops here. So I'm wondering what is the lisp way to handle it. Is it to give a function parameter to the convert-to-cons function which is called for each cons?
<11> I think this is the best way
<1> kingruedi: does "some-code-1" return a list of some unknown length?
<11> no, it just returns something like '((a . b))
<1> so a list with a single element in it? Then you should not use append, but collect instead.
<11> I know. Please ignore that :)
<1> the top-level loop could be rewritten as (loop for i in (convert ...) collect (case ...)))
<1> then, you can turn (case ...) into a function, say fun, that you can call like this: (fun i)
<11> oh, really? I didn't knew that. But that is not my question :)
<1> I know, I am getting to it
<1> if you want help, you have to accept that I give you some other advice at the same time :)
<11> ok :)
<1> now that you have the "fun" function, you can just modify the first function to do collect (fun (some-code-1 ...))
<12> how do i print the contents of list without parentheses showing?
<12> eg (+ 234 344) prints as + 234 344
<5> clhs ~{
<14> http://www.lispworks.com/reference/HyperSpec/Body/22_cgd.htm
<9> marcelino: with (format destination "~{~S~^ ~}" list)
<11> beach: thank you
<7> marcelino: that won't work recursively, though.
<13> kingruedi annotated #23675 with "The updatet function" at http://paste.lisp.org/display/23675#1
<12> i was wanting recursive print out of a tree with indentation to depict hierarchal decomposition
<11> beach: something like this?
<12> similar to indentation that trace would print out
<9> marcelino: I think you might be able to get the pretty-printer to do that with sufficient tinkering of the dispatch table, but I don't know the details
<12> well perhaps the simple approach is to recursively traverse the tree structure e.g. in preorder fashion and just extract nodes and print them
<1> kingruedi: looks plausible. I don't understand why you don't do (loop with ptr = 0 until (>= ptr (length buffer)) as opposed to testing and calling (loop-finish) explicitly
<9> marcelino: that's probably simpler, yes
<11> beach: oh. thank you. I'm not that famillar with loop. until seems to be the better way
<12> it would be nice to have a routine that takes a pretty printed lisp form and prints it with the same indentation level form but without any parens
<15> marcelino: why don't you think two seconds?
<1> kingruedi: you could abstract out (+ (ash (logand ...)) (ash ...) (ash ...)) into a function so as to avoid repeating the code
<15> (delete #\( (delete #\) (with-output-to-string (s) (pprint '(hello (world (+ 1 2))) s))))
<11> beach: ok
<5> pjb: What, no delete-file? ^_-
<15> No this time, two DELETE are enough.
<16> I'd think `prin1-to-string' handy in that case.
<15> Unfortunately, there's no pprint-to-string
<5> Cool! Found a spare 30 gig laptop drive. That should jump my spare machine to actual usable amounts of space.
<17> pjb: is wrapping things in with-output-to-string a major hardship?
<9> (write-to-string x :pretty t :escape t)
<16> kpreid: Missing newline ;)
<9> true, but why would you want it?
<9> there's also no print-to-string
<16> _I_ wouldn't :)
<18> gn8
<5> ... Hunh. I may have been mistaken about some of these Win32 memory functions.
<5> And if I am, then I can build a better posix emulation than cygwin managed. >:-)
<19> what are you working on ?
<5> Right now, background research on a few Win32 APIs.
<7> does anybody else use cl-unification
<7> ? If so, you'll want to fix the MATCH macro, since it generates the initial unification environment at macroexpansion-time.
<5> Hrm...
<5> Okay, I -definately- need to do some experimentation with these functions.
<20> nyef: did you see the recent announcement from MS about being able to download the MSDN Library for free? It's pretty handy to be able to search that stuff locally.
<21> LISP
<5> junrue67: No, I didn't see that announcement. Thank you.
<20> nyef: np
<5> (Not that I have the drive space free for it right now...)
<12> i have a bunch of lisp code in /usr/share/common-lisp /source and systems directory how is this code used say from slime does it have to be compiled with asdf-install before usage?
<22> I'm about to ask onto the mcclim list for this, but thought that someone here might know: Would anyone be aware of any work already made towards the operation of a CLIM-driven analogue to the DESCRIBE operation?
<5> ... You mean like CLIM-LISP:DESCRIBE?
<22> nyef: I was not aware of that function; I will take a look at it; thank you
<22> nyef: well, I was wondering if anyone has set up any parts of a CLIM GUI mechanism, for it. (I know, i could try to work-out a design for such, but maybe someone has already approached it)
<5> What's to do for DESCRIBE, beyond making various bits clicky?
<22> nyef: good point on the clicky-bits part; slime could actually serve to some consideration for design, there
<5> You know what? I've only got about five minutes of runtime left before I go to bed, and I'm not up for trying to figure out where you're going with this.
<22> ok
<22> nyef: appreciationg the conversation though; it's occured to an effect that I've been able to figure out more of what I may need to do for it
<5> ... Okay, I may be tired, but I -know- that sentence wouldn't make any more sense if I wasn't.
<22> fiwI appreciate the conversation.
<22> the conversation occured to an effect: I have been able, now, to have figured out more of what I may ned to do for the design of .. the design of .. of that thing I"m trying to design, I guess
<5> Ah. Glad I could be of some help.
<22> ditto, honestly
<5> ... Yup, I'm crashing.
<5> G'night.
<22> l8r
<23> hi. is there any way to remove the nth item of a sequence?
<24> maybe (remove (elt <sequence> <n>) <sequence> :start <n>)
<25> (remove-if (constantly t) (list 'a 'b 'c 'd) :start 2 :end 3)
<24> and include :count 1
<25> oh, right. :count exists too
<23> jsnell, ewwwww :P
<23> jsnell, well, i'll encapsulate it
Return to
#lisp
or
Go to some related
logs:
dual boot ubutu windows
billiards+ how to rack
Unable to read printer database. Please ensure the foomatic-db package is instal
#php
#ai
fluxbox annoying tabs
22 Balrog
#ubuntu
olivia d'abo bolero underage
gimp rearrange colormap
|
|