| |
| |
| |
|
Page: 1 2 3 4 5
Comments:
<0> And that isn't particularly plausible, either.
<0> (It's probably happening, but I'm unconvinced that it would cause this if it were.)
<1> pebblestone: : I think you can code 2) up by iterating over all symbols, testing if one of the symbol is bound to a generic function and if so, use FIND-METHOD on that to see if it specializes the way you want. There might be an easier way, and it might need the MOP.
<0> Also note that you can dispatch with an EQL specializer. Which is great for singletons.
<2> nyef: just now saw an instance of the GC crash while repeatedly trying to load lisp-unit.lisp. This time, the line number is 4088.
<0> ... Ugh.
<0> This is feeling more and more like a rogue pointer or something.
<1> I wonder how it's possible to check if a symbol is bound to a function in a clean manner.
<0> clhs fboundp
<3> http://www.lispworks.com/reference/HyperSpec/Body/f_fbound.htm
<0> ... Grr... No makefile rule for %.s: %.c?
<1> ah right. so pebblestone, to check if a symbol is bound to a generic function, you can use (and (fboundp 'SYMBOL) (typep (function SYMBOL) 'generic-function))
<4> tcr: have you used cover recently? i'm trying to compile it with sbcl. But it seems to make some troubles
<2> nyef: in what ways could SBCL's state be varying so as to make this so intermittent? As you said before, and in my experience, it happens within seconds of my SBCL session starting.
<5> tcr: that's really helpful, thanks:-)
<2> talking about repeatedly starting SBCL then loading lisp-unit.lisp here.
<4> (okay, it is from 1991 and ansi common lisp is from 1994. That explains a bit :))
<0> junrue67: I have no idea, really. And that worries me.
<0> "movsbl" is -not- a valid x86 instruction.
<6> if they'll have invalid x86 instructions, they should at least have shift-with-exchange as ***
<4> i think GNU as syntax is better. Because movl and movb have different opcodes. So why should they both have one ***embler instruction
<6> Forth ***emblers have done something similar.
<0> ayrnieu: What? *** is -always- Sign EXtend.
<0> (Also known as "CBW", "CWD", and "CDQ".)
<2> nyef: I'm willing to wager a small amount that it's an uninitalized pointer variable.
<0> kingruedi: So sorry. movl %eax, (%ecx) and movl (%ecx), %eax have different opcodes. By your logic, they should have different mnemonics as well.
<2> That would explain a lot.
<2> Or some unitialized variable, anyway.
<7> most of the ***emblers I've written since 1991 have used algebraic syntax
<4> nyef: why not ;) okay, you're right.
<0> kingruedi: Especially given that there are two ways to encode movl %eax, %ecx.
<2> a struct not getting zeroed out, or something like that.
<0> junrue67: An uninitialized pointer -where-? Given that the starting state is deterministic, it would normally get uninitialized to the same value, would it not?
<4> nyef: okay, but actually the x86 encoding is a big problem in my opinion
<2> Do all pages get initialized to a know value (like in a BSS segment) or do you just get whatever was there when pages are mapped?
<2> s/know/known
<0> known value.
<4> tcr: hello? have you any experience with COVER on a common lisp system?
<2> well, it has to be something along those lines. Otherwise, I would think startup is deterministic.
<0> Can you name some of the things which would be different across multiple otherwise-identical startups?
<1> kingruedi: No sorry I don't. I looked around, but couldn't come up with an updated version. I know that it _is_ used by people, though.
<0> (I can think of at least two things which would change.)
<1> kingruedi: so your problems are probably not very severe I reckon.
<2> My steps to reproduce are the same every time.
<2> Well, I mean, I can confidently count on the crash if I follow the same exact steps with enough iterations.
<0> Mmm...
<2> I have various other processes running on my box that are paging stuff in and out.
<2> E.g., iTunes right now.
<0> Time and process ID are the first things that I can think of that would change.
<2> I've debugged time-related problems before, this doesn't have that flavor.
<2> I found a bug in a UUID generator that was due to a bug in processing dates.
<0> You've heard the story of the phase-of-the-moon bug, right?
<2> yeah :-)
<4> tcr: at the moment i'm stuck with the macroexpansion. But i will try searching the web and usenet. I think I'm not that experienced in lisp to solve the problem on my own :/
<0> Okay, now, the core spaces aren't even -mapped- usually, and when memory is ***igned, they're supposed to be 0-fill.
<2> nyef: does the 0-fill happen because of an option p***ed to VirtualAlloc (or whichever the API call is)?
<2> Or because of a zero-ing loop in SBCL code?
<5> tcr: symbol-function is more suitable here, because it takes a symbol as its argument. And the function could be written as:
<5> (defun generic-function? (symbol)
<5> (and (fboundp symbol)
<5> (typep (symbol-function symbol)
<5> 'generic-function)))
<5> oops, sorry
<1> kingruedi: I'm gonna sleep now. I figured you're from Germany as well, so let's put it off.
<2> tcr: before you go...
<2> did that download finish for you?
<2> drat, gone already I guess.
<0> "Memory allocated by this function is automatically initialized to zero, unless MEM_RESET is specified." -- MSDN, on VirtualAlloc.
<1> pebblestone: except for the questionmark, it looks good. :) conventionally, the suffix -p is used in common lisp.
<1> junrue67: Yeah I did, thank you.
<5> tcr: I prefer scheme's naming style.
<5> ;-)
<2> tcr: cool!
<1> junrue67: let's wait for tomorrow, to see how cool it really is :)
<2> tcr: good point!
<2> nyef: is it correct that not *all* of SBCL data is located in these mapped pages? Is there any bootstrap data structure that resides in pages ***ociated with the executable itself, that might not be initialized?
<2> I mean in the pages reserved with VirtualAlloc.
<2> aargh...you know what I mean. There are data pages due to the executable itself (set up by the linker, etc)
<0> The executable image sections are initialized to known values.
<0> (Yes, this includes the bss sections.)
<8> Am having trouble in getting clisp to work with gtk-server via tcp socket.
<2> nyef: ok, enough uninformed questions from me. I will try to work on an easier repro.
<2> bbl
<9> marcelino: Have you gone here and installed the actual gtk-server? the lisp package for gtk-server is AFAIK just a protocol to chat with this program.
<9> marcelino: Here is: http://www.gtk-server.org/index.html
<8> lenolium my gtk-server config seems to work ok because i am able to get it to work using http://leonardoce.interfree.it/gtkserver/documentation.html
<8> which uses stdin
<8> but i am having trouble using tcp socket instead of stdin
<10> marcelino pasted "smart-calc.lisp" at http://paste.lisp.org/display/25670
<8> (defun gtk-start (server nsocket) (ext:run-program server :arguments (list "tcp" (format nil "localhost:~D" nsocket)) :wait nil) (sleep 1) ; wait a little so the server can initialize (socket:socket-connect nsocket)) gets refused.
<9> marcelino: I wish I could help you, but I'm battling with another gtk interface. (clg)
<8> ok no prob.
<0> junrue67: FWIW, the last "really wierd" bug I had with SBCL/Win32 was easy to reproduce... and easy not to. It was an FFI call, and it segfaulted on the return path... unless you wrapped the top-level form invoking it with a LET.
<11> Aargh. Why does latex produce a pdf file instead of a dvi file? My setup must be broken.
<12> rtoym: That could be due to the use of a package with pdf specific options?
<12> I had problems like that before, IIRC
<11> Oh. Could be. How do I disable that?
<11> But that doesn't happen on my (ancient) tex setup at work. Only on my newly installed teTeX on my Mac.
<11> Yay! Finally, an updated cmucl user manual on the web.
<13> hi. does anybody know, whether there is a faculty-function in sbcl? i found ! in clisp. but it doesn't work in sbcl.
<0> "faculty-function"?
<6> nyef - it's like froblitu() in Python.
<0> Okay, what's it like in something I'm familiar with?
<13> faculty 6 = 1 * 1 * 2 * 3 * 4 * 5 * 6
<6> nyef - .liturge in Ruby?
<6> joergen - oh. The factorial function.
<13> aha.
<0> Less familiar with ruby than python.
<11> I think you get to define your own.
<0> In the grand traditions of naming languages after precious stones and snakes, who's writing "diamondback"?
<13> i wanted to use it to define an exponential function using a taylor-polynom. i guess i just try out defining a constant e and using expt instead.
<14> Is opal taken? quartz? citrine?
<6> mbishop - the person who took Pearl and Ruby (no relation) also has Opal, i think.
<11> joergen: What's wrong with (exp x)?
<14> I'm sure Saphire is already taken
<13> ah that's good.
<13> still have to learn a lot.
<11> joergen: A book is usually quite good to help you get started.
<11> Emerald? Topaz? Amythyst?
<13> yeah right.
<14> diamond? leverite (sp?)
<6> anyway, the one true way to define FACTORIAL : (defun factorial (n) (apply #'* (iota n)))
<11> (factorial 1000) probably won't work on something like gcl.
<11> Maybe even 100! won't work.
<15> ayrnieu: typically that's just going to call REDUCE anyway, but you'll end up bound by CALL-ARGUMENTS-LIMIT
<15> of course, maybe that was the joke and I missed it
<11> Apply gets converted to reduce?
<11> chandler: I'm safe from our IT dept, now. :-)
<13> why do i get a floating point underrun calling exp -100 ?
<13> what about zero?
<11> Because you have your FP modes set to trap on underflow.
<11> And (exp -100) gives you single-float precision.
<13> that maybe useful someday.
<11> And exp(-100) is about 3.7e-44, which is smaller than the smallest normalized single-float of 1.1e-38.
<13> kk
<11> Oh cool. I didn't know I could trace labels functions.
<14> hmm
<16> clhs: describe
<3> http://www.lispworks.com/reference/HyperSpec/Body/f_descri.htm
<16> hello.
<17> hi, i using allegro serve and want to intercept request from client and do anything i want
<18> Hey all!
<18> Do you guys know of a good tutorial site / paper I could read?
<18> I've been meaning to learn about lisp (I've always heard there's two types or programming languages, lisp and everything else) and I figured I'd dedicate a few hours tonight to it
<19> unfortunately learning what makes it different will take more than a few hours. you might come away with the wrong impression
<18> wkh: I'm not looking for that, I'm looking to learn lisp
<20> minion: tell robd about pcl
<21> robd: look at pcl: pcl-book: "Practical Common Lisp", an introduction to Common Lisp by Peter Seibel, available at http://www.gigamonkeys.com/book/ and in dead-tree form from Apress (as of 11 April 2005).
<19> well, have you seen Practical Common Lisp?
<19> beaten
Return to
#lisp
or
Go to some related
logs:
#qemu
#perl
can't locate cwd.pm gentoo
disable wireless gentoo
/usr/include/libgdome/gdome.h:27:18: error: glib.h: No such file or directory
#linux
5.0 Fatal error: mysql.user table is damaged or in unsupported 3.20 format
#osdev
xgl on intel915
waysmall startx
|
|