| |
| |
| |
|
Page: 1 2 3 4 5 6 7
Comments:
<0> Neat, slime's arglist for make-instance shows the initargs of the cl*** whose name I've quoted. I never noticed that before.
<1> yeah that's a nice feature
<2> is 'capi' some well-known package ?
<0> capi is the lispworks gui
<3> minion: advice on macroing macros
<4> You can't expect automated advice for everything.
<3> too bad.
<5> minion: advice on expecting automated advice for everything
<4> everything: #11945: Well, what did you expect?
<1> valizas: why do you ask?
<2> interferon: browsing through the different solutions to 2003 ILC contest, I tried to compile a random one out of nine or so, and stopped with error under sbcl
<2> race.lisp
<2> hefner: i see ... i was told it is powerful, so no surprise this contest entry uses it (although I don' t know how was evaluated at the contest)
<1> i would agree with that ***essment
<6> oh geez.
<6> so many problems with cl-sdl
<6> on mac os x
<1> does vim support ***p-based motion commands?
<7> (with-output-to-string (s (make-array 3 :element-type 'character :initial-element #\# :fill-pointer 3)) (prin1-to-string 12))
<7> returns "12"
<7> why not "12#"
<7> ?
<0> several reasons, but note that you are not anywhere writing to the stream 's'
<8> Anyone on familiar with lispworks ide?
<9> good morning
<3> If you can think of better/simpler ways to do templating like <http://paste.lisp.org/display/36596>;, please annotate.
<3> good night
<10> what do you mean by "familiar"?
<2> pkhuong: that seems an astonishing piece of code to me.
<2> pkhuong: where may I read about templating in the sense that you are working on ?
<2> pkhuong: does it have to do with some araneida usage? or is it another meaning for templating ?
<11> how does sbcl's gc record old->new pointers?
<8> Is there a std lisp function that takes one input but always returns nil?
<8> (for ANY input)
<12> does not sound very useful :)
<12> clhs constantly
<13> http://www.lispworks.com/reference/HyperSpec/Body/f_cons_1.htm
<12> there, you have it :)
<12> 2$ from you :)
<14> slava: Forwarding pointers I believe. Do you need the bit pattern?
<11> no, i'm just curious
<11> by forwarding pointers, do you mean a remembered set?
<14> No, when it copies an object from oldspace to newspace it puts a forwarding pointer where it used to live in oldspace. When the next thing that refers to it is copied (or scavanged) it will find the forwarding pointer and know where to readjust to.
<14> I think this is how it works anyway.
<11> oh, we're both confused :)
<11> by old->new i mean pointers from tenured space to the nursery
<11> not oldspace -vs- newspace in the middle of a GC
<14> Tenured space where the write barrier has been hit is scavanged for pointers to newspace, but I don't know the details.
<11> what's the write barrier for slot storage, specifically?
<14> SBCL just uses mprotect for a write barrier, so it's on machine-page boundries.
<11> so when you store a reference to a new object in an old object, the memory protection handler adds the old object to some kind of remembered set?
<14> I think it just marks that page as dirty, and the whole page is scavanged come GC time.
<11> fair enough.
<11> isn't this inefficient, though? i hear a memory protection fault takes up many cycles
<15> hey .. anyone have some suggestions as to how i can call foreign-functions (cffi) with variable number (the count, types and values determined at runtime!) of arguments? ( i've nagged the cll'ers about it and Pillsy came up with a .. hm.. hack .. that works but .. http://groups.google.com/group/comp.lang.lisp/browse_frm/thread/cec20dff595d2859/# )
<14> Yeah, it does. (A few thousand last I checked in Linux.) Once the write barrier is hit for a page it gets unprotected so future writes to that page won't trap. I think most research shows that a good software barrier is faster, but it's also a lot more complicated.
<11> ah
<14> Although I'm not sure whether software barriers being considered faster is primarily because of trap times or because of having better granularity than whatever your machine page size is.
<14> Note that decent kernel support for reading dirty page marks would eliminate the stupid trap overhead - I think you can actually do this in Win32.
<11> indeed
<11> but doesn't the kernel use the dirty bit for swapping?
<11> so if the write barrier is hit, the page is swapped out, and swapped back in, it won't be set anymore?
<14> It could keep an application-visible copy.
<14> That's only reset when the app asks.
<14> (Perhaps that's what win32 does, I don't know...)
<2> clhs prog2
<13> http://www.lispworks.com/reference/HyperSpec/Body/m_prog1c.htm
<2> "prog2 evaluates first-form, then second-form, and then forms, yielding as its only value the primary value yielded by first-form"
<11> that's a well-known typo
<2> thanks
<11> the correct behavior is (prog2 1 2 3) ==> 2
<2> should say second-form right ?
<11> right
<11> otherwise, prog1 and prog2 are the same function
<2> i see. are there other well-known typos in clhs ?
<11> i forget the details, but as specified, subtypep can only be implemented if you solve the halting problem...
<11> clhs subtypep
<13> http://www.lispworks.com/reference/HyperSpec/Body/f_subtpp.htm
<11> or maybe it was upgraded-array-element-type
<16> slava: Why would subtypep be non-terminating?
<11> i don't remember exactly, but the specification of either subtypep or u-a-e-t had a problem with 'satisfies' types
<11> ask Xof when he emerges from cryogenic stasis
<17> josh: satisfies
<17> anyhow, the only thing missing from subtypep is an oracle, which could do it.
<18> She says it looks like I'm waiting for something. :(
<19> hmm, is there a *nix common lisp interpreter shell with command history and preferably highlighting, or should I just bother with learning emacs? :)
<17> well, clisp has a readline interface ...
<15> no point in wasting time .. go for emacs+slime
<15> :)
<20> The CLIM listener has syntax highlighting and history.
<19> thanks for the suggestions, I'll try :)
<20> (You still want to learn Emacs, but as the CLIM listener gets more stable, it's going to be an interesting alternative to the normal REPL, I believe.)
<19> yeah, emacs is probably very nice for lisp, it's just that I'm so used to vi learning emacs commands takes a good while :)
<14> It's almost certainly worth your time to learn emacs commands anyway, but there's also M-x viper-mode, a vi (not vim) command emulator for emacs.
<19> maybe. is it possible to run a lisp interpreter in an emacs window so that it evaluates it automatically
<19> the *scratch* buffer hints that it's for Lisp evaluation but I don't know how to evaluate it heh :)
<14> Yes, see Slime.
<14> minion: slime?
<4> slime: SLIME is the Superior Lisp Interaction Mode for Emacs. http://www.cliki.net/slime
<19> hmm, ctrl-x ctrl-e seemed to evaluate it :)
<19> ok thanks
<14> Yes, but that's evaluating it in emacs-lisp, which is not Common Lisp and is generally not recommended for anything but programming emacs.
<19> thought so..
<2> bdowning: is it not recommended for programming emacs for safety reasons ? style?
<2> bdowning: sorry, misread
<19> slime seems worth learning emacs
<21> it is!
<22> Moin
<23> <11>i don't remember exactly, but the specification of either subtypep or u-a-e-t had a problem with 'satisfies' types
<23> (upgraded-array-element-type '(satisfies arbitraryp)) is defined by ANSI to return the "right" answer
<23> (which it can't possibly do)
<23> in general, there's a whole list of noted bogosities in the standard on Cliki
<23> see <http://www.cliki.net/Proposed%20ANSI%20Revisions%20and%20Clarifications>;
<19> hm...how could I tell slime to print its output of the lisp evaluation to a separate emacs window, now it just display 1 line of the result in the echo area...
<21> joga: by pasting it in the repl buffer
<21> lisppaste: paste 27885?
<21> aha (:
<21> well, I use this: http://paste.lisp.org/display/27885
<19> ah so I'd just need to mark the part I want evaluated and copy&paste
<21> use M-x asf-slime-copy-defun-to-repl RET (or bind it to a key), and you get the current top-level form in the repl buffer. it should work like C-c C-c or C-M-x
<21> oooh
<23> I'm slightly surprised to find that that isn't that part of slime
<21> oops, there is something more similar to what joga describes!
<23> (I use C-c C-c to do what joga wants, but of course it only prints the output, not the input
<24> Would be kind of useful if C-u made C-M-x paste the result into the buffer though, just like it does for `shell-command'.
<21> C-c C-p, slime-pprint-eval-last-expression
<21> works like C-x C-e
<21> lichtblau: or like the same command in emacs-lisp-mode
<20> Hmm, C-u C-M-x in emacs lisp is something else. "With a prefix argument, instrument the code for Edebug."
<25> morning
<21> oh, right
<21> _8work: I was thinking of C-x C-e
<26> While dumping a list with (format t "~{...~}" lst) what's the most convenient way to enumerate the displayed items?
<27> what do you mean by that?
<26> Xach: I want No. 1
<26> No. 2 in the (format t "~{~{~a:~10t~a~%~}~%~}" *db*)) output.
<27> ah
<27> hmm. dunno.
<26> :)
<27> i'd probably use LOOP in that situation.
<26> Maybe there's a way similar to (format t "~{No. ~2t~a:~%~{~a:~10t~a~%~}~%~}" (iota (length *db*)) *db*))
<27> that way ****s
<28> What does iota do?
<27> generates a list of numbers
<26> asbjxrn: Umm... a SRFI function from Scheme. (iota 5) -> (1 2 3 4 5)
<26> I'm not sure if it exists in Lisp.
<27> traversal and consing make it a big bummer for this.
<27> it doesn't, but it's trivial to write.
<26> Yep.
<29> why is it called iota?
Return to
#lisp
or
Go to some related
logs:
gentoo getting rid of kde-env
#sed
installing gazpacho on cygwin
You must have the XTest library to build synergy
BIN_FILE bash
gnome swithdesk fc3
make-kpkg --arch xen
WEXT auth param 7 value 0x1 - ioctl[SIOCSIWENCODEEXT]
ghostscript rpl_malloc
Bareword gaim not allowed
|
|