freenode/#lisp - IRC Chatlog

2:46:19 loke The defmethod macro itself calls ENSURE_GENERIC-FUNCTION... The defmethod macro expands into a call to INSTALL-METHOD with also calls ENSURE-GENERIC-FUNCTION.

2:46:23 loke So there are two calls?

2:47:03 loke I'm looking at the ECL code, not Clasp.

2:58:57 drmeister There may be - I've got problems much later - when methods are being defined and wiping out the specializer-profile slot that I added.

3:00:35 drmeister I can't seem to find the point where the lambda-list is defined and the generic-function is initialized for the first time.

3:03:56 loke Isn't that in ENSURE-GENERIC-FUNCTION? There is a cond clause for (not (fboundp name))

3:04:56 drmeister The lambda-list doesn't appear to be defined in some cases - although I may have done something wrong.

4:51:41 JohnTalent Bueller?... Bueller?... anyone?... Bueller?

5:00:51 lexicall hi, I wonder is there an elegant way to call a method of super class? In order to do that I have to use CHANGE-CLASS but I think it's quite ugly.

5:02:52 loke lexicall: (call-next-method)

5:04:03 lexicall thanks!

5:19:44 jasom did sbcl core images get a lot smaller recently?

5:20:18 jasom I'm seeing 44MB uncompressed for a fairly simple image on x86_64, and I rememberd that number being higher...

5:37:12 beach Good morning everyone!

5:37:26 loke Hello beach

5:37:31 loke ACTION time for lunch

5:43:27 iqubic I hate still being banned from #emacs

5:43:45 iqubic It been a week and a half now.

5:51:40 scottj iqubic: fwiw when it happened I think they said a 24 hour cooling off period so maybe message the op that banned you.

5:52:01 iqubic I don't remember who banned me.

5:55:00 pillton jasom: I don't think so. http://paste.lisp.org/display/355631

5:59:03 pillton Well I guess 6M.

6:13:25 Shinmera jasom: Stassats has been continuously working on reducing core sizes.

7:05:39 temporal1 ** NICK TeMPOraL

7:15:43 Kovalski ** NICK neko

7:16:14 neko ** NICK Guest4586

7:16:16 hajovonta hi everybody

7:16:22 Guest4586 ** NICK Kovalski

7:16:38 Kovalski ** NICK nulani

8:00:11 shka hello

8:00:46 beach Hello shka.

8:00:59 shka question regarding non standard generic function metaclass

8:02:07 beach Yes?

8:03:13 shka will performance be hurt under sbcl even if i won't use non-standard dispatch?

8:03:50 beach I don't think so.

8:04:20 shka hm

8:04:30 shka well, i guess we should check

8:04:50 beach What does your non-standard generic function class look like?

8:05:46 shka like here

8:05:51 shka ah, my

8:06:33 shka beach: no members, no specializing CL GFs

8:06:47 shka i need it just to add my own methods

8:06:51 shka to my own GF

8:14:57 shka beach: i will show code once i will write it

8:15:02 TMA GF abbreviation is very context dependent... it means girlfriend outside of lisp related chat

8:15:17 shka TMA: girlfriend? What is that?

8:15:39 hajovonta :)

8:17:38 Shinmera And in yet other contexts it means Galois Field

8:18:16 TMA a much more approachable concept

8:18:30 shka indeed -_-'

8:31:10 JuanDaugherty grammatical framework to me

8:31:44 JuanDaugherty (for which there's a channel here)

8:33:30 shka JuanDaugherty: btw, your nick

8:33:37 shka Joni Mitchell?

8:33:52 JuanDaugherty it's on my birth certificat

8:33:56 JuanDaugherty e

8:34:07 shka oh, ok

8:36:02 JuanDaugherty but take it you are referring to Don Juan's Reckless Daughter

8:36:09 shka yup

8:36:58 JuanDaugherty the Irish name Daugherty doesn't actually come from daughter, just sounds like it does

8:37:13 shka i guess so

8:47:16 shrdlu68 Good morning, Commoners.

8:51:25 phoe__ hey shrdlu68

8:52:54 hajovonta hali

9:21:37 Posterdati hi

9:21:50 Posterdati shrdlu68: commodores please!

9:23:59 White_Flame shrdlu64

9:26:18 Posterdati yes, seems resonable

9:52:07 Posterdati please, is there a way to trigget a call to the io-handler in the iolib multiplexer?

9:52:25 pjb Can't you call it directly?

9:53:49 Posterdati pbj: I do not know how to funcall it

9:54:11 Posterdati it is set by set-io-handler

9:55:54 pjb (set-io-handler #'my-nice-handler) (my-nice-handler) ?

9:57:40 Posterdati pjb: I have no access to my-nice-handler

9:59:05 Posterdati pjb: I modified the ex8-server code in iolib sockets example

9:59:06 pjb Why do you want to do that?

9:59:17 pjb calling the io-hander?

9:59:29 Posterdati pjb: because I do not need an echo server!

10:01:05 pjb I don't see the relationship between "you do not need an echo server!" and "trigger a call to the io-handler".

10:01:51 Posterdati pjb: I'm modifying the echo server example

10:02:20 Posterdati pjb: because a I need a server capable to get some commands over a socket and then answer back

10:02:21 pjb (iomux:set-io-handler eb socket :read (setf *saved-handler* (make-whatever-handler-you-want socket id etc))) (funcall *saved-handler* fd event exception)

10:02:43 pjb Again, I don't know why you would want to do (funcall *saved-handler* fd event exception)

10:03:06 pjb If there's no I/O why would you want to call it? If there is I/O then it's already called by iolib!

10:03:07 Posterdati because I want to write out data to client

10:03:13 pjb Then write data!

10:03:24 pjb (write-sequence data socket)

10:03:44 Posterdati not using a multiplexer io handler?

10:04:16 pjb The multiplexer is for asynchronous I/O.

10:04:43 pjb otherwise check ICMP.

10:37:54 hajovonta guys are you aware of a CL library that can do asynchronous ICMP echoes? I need to ping a host periodically and I couldn't find one. Is it a thing that is practically done via FFI?

10:41:19 pillton All I/O is done by FFI.

10:43:54 loke To my knowledge there is no FFI library to do direct IP

10:44:28 fe[nl]ix iolib has an example of how to create ICMP packet

10:44:30 fe[nl]ix s

10:45:13 loke wait what? IOLIB can do it?

10:45:38 fe[nl]ix but you most likely don't want to do that because creating raw sockets requires basically root privileges

10:45:45 fe[nl]ix at least on Linux

10:46:21 fe[nl]ix you're better off spawning a thread that runs /bin/ping and reads its stdout

10:50:34 fe[nl]ix loke: https://github.com/sionescu/iolib/blob/master/extras/ping.lisp

10:51:17 loke ice

10:51:20 loke nice

10:51:49 hajovonta hm

10:51:52 hajovonta am I?

10:51:56 hajovonta that, I can do.

10:52:31 hajovonta I'm just a little worried that different linux ping versions may print different output

10:52:41 fe[nl]ix a more sophisticated approach would be to make a small C binary that creates a raw socket and sends it to the main process via a Unix socket

10:52:50 fe[nl]ix then you can setuid that binary

10:53:42 hajovonta I've already seen the example you wrote above but wasn't able to make it work

10:54:13 hajovonta I don't have C knowledge

10:54:15 fe[nl]ix iolib already has code for receiving file descriptors via a Unix socket, but you'd have to write the C to create and send the file descriptor

10:54:45 fe[nl]ix hajovonta: when I wrote that code I was running Gentoo

10:55:09 fe[nl]ix and I modified the kernel to not require root for raw sockets

10:55:17 hajovonta I see :)

10:55:21 fe[nl]ix recompiling your kernel is easy on Gentoo

10:55:40 shrdlu68 It's routine.

10:56:02 hajovonta everything is routine when you've done it enough times :)

10:56:29 fe[nl]ix hajovonta: the output of /bin/ping can vary between OSes but it should be pretty stable on the same OS

10:57:48 hajovonta that's true. The output is not standardized

10:57:51 fe[nl]ix you can safely assume that /bin/ping is the GNU variant

10:57:57 fe[nl]ix on Linux

10:58:05 hajovonta also, the switches of /bin/ping can differ

10:58:40 shrdlu68 How about using iolib and setting suid?

10:58:50 fe[nl]ix on weird distros it could be a symlink to Busybox, but I doubt you care about those

10:58:53 shrdlu68 ping essentially does that...

10:59:31 fe[nl]ix shrdlu68: sbcl re-executes itself on start and I don't think that setuid carries along

10:59:37 hajovonta or maybe I can write a python lib that uses raw icmp and uses standardized output and just call that. ;)

10:59:59 hajovonta or C

11:00:07 fe[nl]ix also, if you save a new executable image you have to setuid it again, etc...

11:00:09 hajovonta but I don't know C, so...

11:00:37 fe[nl]ix look at this as a good opportunity to learn C

11:00:51 fe[nl]ix you have a specific goal, which is very good

11:00:58 fe[nl]ix instead of generically "I want to learn C"

11:01:14 hajovonta I already got in touch with C several decades ago...

11:01:23 shrdlu68 fe[nl]ix: I can't confirm, have you tried it?

11:01:28 d4ryus4 ** NICK d4ryus

11:01:31 hajovonta the same is true for Java, PHP and Perl :)

11:01:36 fe[nl]ix shrdlu68: in the past, yes

11:02:26 shrdlu68 Well he could use a different lisp...

11:04:24 hajovonta for example?

11:04:33 shrdlu68 ecl, ccl

11:05:44 shrdlu68 If the problem here is that one can't effectively use setuid with sbcl, there are other implementations that hopefully don't have this limitation.

11:12:25 _death you could just use setcap instead of the suid bit

13:17:23 francogrex hi is it possible to reconnect to a running thread: debugger invoked on a SB-INT:SIMPLE-READER-PACKAGE-ERROR in thread #<THREAD "main thread" RUNNING {23EE7E89}> sbcl hangs and i would like to open another sbcl and "attach" it to that hanging running thread

13:23:34 antoszka otwieracz: ^ haven't you been dealing with a similar problem recently?

13:34:25 otwieracz Nope.

13:36:52 francogrex quit

14:15:41 hjudt can anyone help me with cl-json? i want to generate json output like this: { headers: ["h1", "h2", "H3"], data: ["d1", "d2", "d3"] }. i have headers and data as lists. what do i need to do?

14:16:33 Bike according to the manual, you can get a js object if you feed it an alist or a hash table.

14:16:35 hjudt simply (list headers data) gives indexed arrays, but i'd like to have named arrays

14:16:48 Bike so you could do like (list (cons 'headers headers) (cons 'data data))

14:18:17 hjudt (cl-json:encode-json-to-string (list (cons 'headers '(h1 h2 h3)) (cons 'data '(d1 d2 d3)))) => "[[\"headers\",\"h1\",\"h2\",\"h3\"],[\"data\",\"d1\",\"d2\",\"d3\"]]"

14:18:51 hjudt not what i want. i'd need "[ headers: ..., data: ...]"

14:19:50 Bike can't believe the manual would deceive me.

14:20:28 Bike oh, i guess it's because they're lists. genius.

14:21:08 Bike okay, if you do a hash table it does work properly.

14:21:27 Bike as in, i get {\"headers\" ... } out.

14:21:35 scymtym (json:encode-json-alist '(("headers" . ("h1" "h2")) ("data" . ("d1" "d1")))) should also work

14:21:54 Bike oh, a trick. nice

14:22:13 hjudt scymtym: thanks. that's the solution.

14:35:26 capitaomorte ACTION alias

14:35:48 capitaomorte (sorry, typo :-)