| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10
Comments:
<0> because it's an arbitrary number of values
<0> (values simple-string &optional) is exactly one value
<1> Xophe: I see. Why is my (DECLARE (VALEUS SIMPLE-STRING)) being transformed into that, however?
<0> because (values simple-string) means (values simple-string &rest t), according to ANSI
<2> what is this split between gcl 2.6 and 2.7? On debian, 2.6 seems maintained (but lacking features) and 2.7 unmaintained (though this might just be lack of a rebuild from unofficial upstream cvs)
<0> (also, (declare values) is a non-standard declaration which might go away)
<2> Xophe: would it be replaced by something?
<2> Xophe: or only by a declare ftype in the outer scope?
<1> Xophe: How can it be "according to ANSI" and "non-standard"?
<0> possibly, though I'd like to know what it can do that can't be done with either an ftype declaration or by having a (the ...) around your return value
<0> quotemstr: the type (values simple-string), according to ANSI, ... ; and the declaration (declare values) is non-standard
<0> since you managed to typo what you said you had, I couldn't tell what you actually meant
<1> Okay.
<1> Yes.
<3> Fare: I think gcl 2.7 is the development version of gcl and still in progress. Dunno about debian.
<2> the gcl 2.7 in debian seems to have issues with eval-when.
<0> Fare: have you tried reporting that as a bug to the people who want to hear about bugs?
<0> (in this case, to the very fine Debian bug tracker)
<2> Xophe: I filed a grave bug report for gclcvs on debian
<2> for failing to install properly
<4> Fare: I just read your matrix-variant implementation of the fibonacci sequence. While I had seen the trick in some haskell code it was a nice read
<2> jlouis: thanks :)
<2> (I'm vaguely hoping that if the debian package is updated with latest CVS, the eval-when problem might go away)
<0> you know that you can try building it yourself to see if your vague hope is to be joyfully realised?
<5> Fare - parenthesized remarks don't come across well in Lisp channels.
<1> How do FTYPE declarations interact with ordinary declarations of argument types to functions?
<6> good afternoon. I haven't seen the sbcl win32 TODO list in a while. But, does anyone know if sbcl on win32 supports loading of shared object files? If so, I ***ume it takes windows .dll files, no cygwin .so files
<7> Cygwin does .so files?
<7> (Yes, it loads .dll files.)
<6> nyef: I thought so. but, I've actually only seen .a files from cygwin
<6> nyef: okay, that's helpful. I'll try some more with .dll files and see if I can get success. thanks for the confirmation
<7> kmr: You might try finding some example code which is known to work.
<6> nyef: I have .dll files for clsql that work on clisp. That's what I was trying with sbcl.
<7> You might take a look at http://www.lisphacker.com/blog/display/some-sbclwin32-hacking which links a tarball which should show the technique.
<6> nyef: thanks for the tip!
<0> rtoy_: with debian it's relatively straightforward
<0> apt-get source gclcvs; also, pull down gcl cvs. Apply the debian patch, dpkg-buildpackage -us -uc -rfakeroot -b
<3> It would be easy if I had a debian system. :-) (Suse 9.0)
<6> rtoy_: hi. perhaps install a debian system in chroot of your suse system?
<4> I have a Xen box for such experiments
<3> kmr: If I had enough space, I might just install a whole version, but I don't. And I'm reluctant to replace my Suse 9.0 system for something newer lest I break something and can't build cmucl/x86 anymore. :-)
<6> jlouis: that's reasonable and i do that on some boxes. but, installing a xen kernel and consifuring the userland can be difficult
<6> rtoy_: sorry to hear about the space crunch. you'd probably like to have 2GB for a fairly complete text-mode development system.
<6> rtoy_: I'm sure your familiar with the trick creating a large file with dd, then attaching it to a loopback device, then mkfs and mounting the loopback device
<3> kmr: Mostly laziness and lack of available space on the primary disk to install a small root partition from which to boot.
<4> kmr: indeed!
<3> I think I have space to add another drive, but I think that would mess up the existing partitions holding linux, so I've been reluctant to do that either.
<6> don't need to change the partitioning. in fact to get quickly up and going, you can likely find a sarge filesystem image you can download and just mount
<3> Yeah, any live linux CD would work too, I guess. But I'm kind of holding out for a bit and getting perhaps a new box since I can no longer even install any version of Suse later than 9.0 on my box. The install CDs won't even boot.
<3> Now it just as small matter of convincing the boss at home that it would be useful.
<3> it's
<6> ah, the damnable cost/benefit analysis
<6> rtoy_: yeah, a live cd would work. also, a xen virual filesystem from http://jailtime.org would work
<6> I'm using centos4.3 that I downloaded from jailtime.org under xen to run brightmail anti-spam on a debian box. works well.
<6> rtoy_: btw, I'm looking at taking some of your cmucl code (src/loop.lisp) to use in clsql with clisp since clisp doesn't have an extensible loop macro
<6> rtoy_: would you have time to take a look at a #+cmu form in the file for your porting thoughts?
<3> Because I can't install Suse 9.x, I wonder if the newer live cd's will even boot on my system. That's extra motivation for a new box. :-)
<3> kmr: No problem taking anything from loop.lisp. My changes are free, and loop.lisp is free. I can't look at it right now, but if you send me a memo I'll look at it tonight or tomorrow.
<6> rtoy_: I was thinking rather than booting a live cd, of mounting a live partition under suse and chroot'ing into that partition. but, there's a good chance that the live partition is created dynamically from a compressed form on the cd
<6> rtoy_: thanks, I'll drop you an email
<6> if you have a 1GB to spare, probably the fastest way to have a debian system available is to download the sarge image from jailtime
<6> http://jailtime.org/download:debian:v3.1 -- 53MB download
<6> probably expands into a 512-1024MB partition
<6> s/partition/filesystem/
<3> kmr: Thanks for the hints. I'm not really in a hurry to have a debian system. My Suse system works fine for what I use it for these days, which is build cmucl.
<3> Oh, gotta go. bbl, perhaps.
<6> very fine. bye!
<2> woohoo, wish granted: camm maguire just updated the gcl 2.7.0 deb
<8> pjb: Why not (tor maybe t) => T?
<9> Yes, that could be.
<9> You have to choose one logic.
<9> Yes, (tor maybe t) = t is probably better, since I have (tand maybe nil) = nil
<10> definitely makes more sense.
<8> I love the twhile cond case for maybe
<8> Out of curiosity, what do you use this for personally, pjb?
<9> Nothing yet.
<9> but it could be used with a sql db :-)
<1> What on earth does "cannot use values types here" mean?
<7> quotemstr: What do you think it means?
<1> nyef: Well, I'm trying to compile a function that uses an alien function, and something in the alien internals (some type translator function) is signaling that error.
<7> So... it's saying that you screwed up the alien type definitions?
<8> maybe it means that an alien can't accept multiple values?
<1> Yes, but there doesn't seem to be anything strange about this type definition.
<1> It's (function int (* int) (* char) int (* void) unsigned-long (* char)
<1> ).
<7> Why don't you lisppaste the entire function?
<1> Ah, there we go. (* void) should be (* t).
<11> someone might want to make http://sbcl.org/ work like http://www.sbcl.org/
<6> _jpl_: I don't think that sure that sourceforge's vhosting can do (at least at the time I setup the dns records for sbcl.org). Check "dig sbcl.org"
<12> Hi, I'm trying to run cl-opengl. When loading it, I get a bunch of Undefined Aliend warnings (on SBCL 0.9.15). My guess is that I am using the wrong version of libGL. Does anybody happen to know which is the right one?
<13> looks like it will try any of libGL, libGL.so.1 or libGL.so.2
<13> what OS are you using?
<13> and have you installed current drivers for your graphics card?
<14> kmr: www.sbcl.org points to 66.35.250.210, whereas sbcl.org points to 66.35.250.120. I wouldn't expect that to work. (There may, as you say, still be vhosting problems as well.)
<15> could still be a redirect
<14> You can't connect to port 80 on that address.
<16> does montezuma have a dev list? (defaults to no)
<14> Also, telnetting to vhost.sourceforge.net port 80 and giving "Host: sbcl.org" gets the SBCL web page, so fixing the DNS record should make it work.
<15> bdowning: I mean, is there any reason there couldn't be sbcl.org. A <some machine which can run a redirecting server>?
<17> lol, I love stale fasl's
<14> kpreid: No reason why not, but why bother if pointing it right at sf.net's vhosting server would do the job. (Given that 1 and 2 are transposed that's what I'd guess tried to happen.)
<15> bdowning: true, that's simpler if possible
<14> It works, I just tried it. (See my note about p***ing "Host: sbclk.org" in via telnet.)
<14> sbcl.org
<6> bdowning: www.sbcl.org is a CNAME for vhost.sourceforge.net. I just updated the dns records for sbcl.org to be the same CNAME
<14> Awesomeness. Thanks kmr!
<6> bdowning: glad to, thanks for your help
<1> Is the return value of SYMBOL-NAME safe to modify with nsubstitute?
<14> I doubt it - that sounds like nasal demon territory.
<1> Ok.
<14> (Specifically, the package hash table wouldn't have the right entry anymore if it's interned.)
<14> clhs replace
<18> http://www.lispworks.com/reference/HyperSpec/Body/f_replac.htm
<14> no
<14> sorry
<14> clhs substitute
<18> http://www.lispworks.com/reference/HyperSpec/Body/f_sbs_s.htm
<19> or mapcan + coerce :\
<1> Substitute works, but I can only use one character as a replacement.
<14> Oh, I didn't see "one or more"
<1> I can split the string out into a list of one-character things are concatenate them back again, but that seems wasteful.
<14> I'd probably use an iteration-construct-of-choice and print the output to a WITH-OUTPUT-TO-STRING stream.
<1> Hrm, that works. Thanks.
<19> yeah, w-o-t-s sounds best. Or (apply 'concatenate 'string (map 'string (lambda (char)) string)) if you wanna go functionnalish :)
<13> (loop for char across string for index from 0 do (setf (aref string index) (some-function-of char))?
<14> And run afoul of CALL-ARGUMENTS-LIMIT. :)
<13> nevermind, misinterpreted the one or more part
<12> _3b: I am on linux, (debian testing). I am going to check the driver version. Thanks.
<20> I'd like to use Scheme also, is there something SLIME-like for it?
<19> bdowning: ok, reduce concatenate :p
<14> There's slime48, but that's only for scheme48 I think.
<12> _3b: Well, actually libGL.so and libGL.so.1 are links to libGL.so.2
<1> call-arguments-limit = 536870911
<1> :-)
<13> nowhere_man: might try quack, don't remember how it compares to slime though
<14> You could probably also do something with Series. (I know nothing about series.)
<1> Yeah, I've been looking at that.
<13> mugrido: what graphics card do you have?
<19> bdowning: I wonder what sequence comprehension could be like in a CL-ish package.
<19> Most of the fusion rules can be applied locally too.
<9> quotemstr: (with-output-to-string(out) (loop for ch across string do (princ (something ch) out)))
<9> /clear can do.
<1> TIMTOWTDI. :-)
<1> bdowning: Why restart Emacs? Just clear the buffer.
<19> quotemstr: but usually one of them's better ;)
<7> bdowning: Just clearing the buffer should suffice.
<14> /clear seems to be a no-op to me.
Return to
#lisp
or
Go to some related
logs:
Call to undefined function ssh2_connect() PHP5
libQtGui_debug gentoo
ubuntu freshclam can't parse the config file
xgl RS200
#gentoo
#physics
iptables snat example tap0
sux X11 connection rejected because of wrong authentication
#debian
+cdrecord has no permission to open this device +ubuntu
|
|