December 2004 - slime-devel - mailman3.common-lisp.net

[slime-devel] Avoid low apply limit
by Eduardo Muñoz 11 Jan '05

11 Jan '05

LispWorks has a low args limit for apply. The following patch (inlined below) avoids landing on sldb when inspecting long lists. I think that its safe to #'nconc (nreverse ... but there isnt a noticeable change of speed or consing so it may be more conservative to just use reduce. Index: swank.lisp =================================================================== RCS file: /project/slime/cvsroot/slime/swank.lisp,v retrieving revision 1.272 diff -u -r1.272 swank.lisp --- swank.lisp 16 Dec 2004 21:16:50 -0000 1.272 +++ swank.lisp 22 Dec 2004 21:56:41 -0000 @@ -2631,7 +2631,7 @@ (let ((a (if (null l) a (cons (label-value-line :tail l) a)))) - (apply #'append (reverse a))))))) + (reduce #'nconc (nreverse a))))))) (values title (append '("Elements:" (:newline)) lines))))) (cond ((not length) ; circular -- Eduardo Muñoz | (prog () 10 (print "Hello world!") http://213.97.131.125/ | 20 (go 10))

5 6

[slime-devel] unix domain sockets
by Marcus Breiing 10 Jan '05

10 Jan '05

About two weeks ago I asked on IRC whether there was support for using SLIME via unix domains sockets. Someone answererd in the negative, citing lack of unix domain socket support on Emacs as the reason. I later realized that this doesn't really matter, because we should be able to use an arbitrary process object with a suitable external command. I tried it, and it's working very nicely so far, using unixclient path-to-socket ucspi-proxy as a command in a process object on the Emacs side ("unixclient" and "ucspi-proxy" are available in Debian in the ucspi-unix and ucspi-proxy packages.) (A side effect is that we can easily add other components into the communications pipeline. Things I tried successfully: recordio 2>/tmp/slime-io unixclient path-to-socket ucspi-proxy to record communications, and ssh remote-host unixclient path-to-socket ucspi-proxy to use it from a remote machine.) Below is what I hacked into swank.lisp in order to make unix domain sockets work. Instead of adding specific code to create unix domain sockets, I changed setup-server to accept a pre-established socket object. This is a simple change that should allow anyone on a Lisp that supports it to use unix domain sockets without much change to the swank proper. For the Emacs side, I don't have anything resembling useable patches (being a total slime newbie, really). Something along the lines of the above would probably be simplest, i.e. passing an established process object for slime to use. Some necessary changes identified so far seem to be: * changing (eq (process-status conn) 'open) to (memq (process-status conn) '(run open)) in a few places * Avoid trying to get the "port" of a non-network process object. I need to set (setf swank:*use-dedicated-output-stream* nil) to make things work properly. Interaction looks to be fast enough, though. Marcus diff -u slime-1.0-original/swank.lisp slime-1.0/swank.lisp --- slime-1.0-original/swank.lisp Sat Sep 4 18:19:52 2004 +++ slime-1.0/swank.lisp Sat Dec 18 14:09:36 2004 @@ -277,8 +277,10 @@ dont-close) "Start the server and write the listen port number to PORT-FILE. This is the entry point for Emacs." - (setup-server 0 (lambda (port) (announce-server-port port-file port)) - style dont-close)) + (create-swank-server 0 style + (lambda (port) + (announce-server-port port-file port)) + dont-close)) (defun create-server (&key (port default-server-port) (style *communication-style*) @@ -286,37 +288,37 @@ "Start a SWANK server on PORT running in STYLE. If DONT-CLOSE is true then the listen socket will accept multiple connections, otherwise it will be closed after the first." - (setup-server port #'simple-announce-function style dont-close)) + (create-swank-server port style #'simple-announce-function dont-close)) (defun create-swank-server (&optional (port default-server-port) (style *communication-style*) (announce-fn #'simple-announce-function) dont-close) - (setup-server port announce-fn style dont-close)) - -(defparameter *loopback-interface* "127.0.0.1") - -(defun setup-server (port announce-fn style dont-close) - (declare (type function announce-fn)) (let* ((socket (create-socket *loopback-interface* port)) (port (local-port socket))) (funcall announce-fn port) - (ecase style - (:spawn - (spawn (lambda () - (loop do (serve-connection socket :spawn dont-close) - while dont-close)) - :name "Swank")) - ((:fd-handler :sigio) - (add-fd-handler socket - (lambda () - (serve-connection socket style dont-close)))) - ((nil) - (unwind-protect - (loop do (serve-connection socket style dont-close) - while dont-close) - (close-socket socket)))) + (setup-server socket style dont-close) port)) + +(defparameter *loopback-interface* "127.0.0.1") + +(defun setup-server (socket style dont-close) + (declare (type function announce-fn)) + (ecase style + (:spawn + (spawn (lambda () + (loop do (serve-connection socket :spawn dont-close) + while dont-close)) + :name "Swank")) + ((:fd-handler :sigio) + (add-fd-handler socket + (lambda () + (serve-connection socket style dont-close)))) + ((nil) + (unwind-protect + (loop do (serve-connection socket style dont-close) + while dont-close) + (close-socket socket))))) (defun serve-connection (socket style dont-close) (let ((client (accept-connection socket)))

3 3

[slime-devel] finding setf methods
by Lynn Quam 31 Dec '04

31 Dec '04

Is there any way to do slime-edit-definition on setf methods or setf functions?

2 1

[slime-devel] Problem with Swank and SBCL Unicode support?
by Dave Roberts 29 Dec '04

29 Dec '04

Just had SLIME explode quite magnificently for me. ;-) Was playing around with TRIVIAL-SOCKETS on SBCL 0.8.17.24. You'll need these simple definitions, in my case in foo.lisp: (defun crlf (s) (write-char #\Return s) (write-char #\Newline s)) (defun http-get (site page) (with-open-stream (s (trivial-sockets:open-stream site 80)) (format s "GET ~A HTTP/1.0" page) (crlf s) (format s "Host: ~A" site) (crlf s) (crlf s) (force-output s) (loop for line = (read-line s nil nil) while line do (format t "~A~%" line)))) Test case is pretty simple. Start SLIME. In the SLIME REPL: CL-USER> (require :asdf) NIL CL-USER> (require :trivial-sockets) NIL CL-USER> (load "/home/dave/foo.lisp") T CL-USER> (http-get "www.cboe.com" "/") <lots of output...> <and then...> debugger invoked on a TYPE-ERROR in thread 14342: The value 956 is not of type (UNSIGNED-BYTE 8). You can type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL. restarts (invokable by number or by possibly-abbreviated name): 0: [ABORT ] Abort handling SLIME request. 1: Reduce debugger level (leaving debugger, returning to toplevel). 2: [TOPLEVEL] Restart at toplevel READ/EVAL/PRINT loop. (SB-IMPL::OUTPUT-BYTES/LATIN-1 5 #<FILE-STREAM for "a constant string" {9829B21}> "(:debug 0 2 (\"The value 956 is not of type (UNSIGNED-BYTE 8).\" \" [Condition of type TYPE-ERROR]\" nil nil) ((\"ABORT\" \"Abort handling SLIME request.\") (\"ABORT\" \"Reduce debugger level (leaving debugger, returning to toplevel).\") (\"TOPLEVEL \" \"Restart at toplevel READ/EVAL/PRINT loop.\")) ((0 \"(SB- IMPL::OUTPUT-BYTES/LATIN-1 5 #<FILE-STREAM for \\\"a constant string\\ \" {9829B21}> \\\"(:debug 0 1 (\\\\\\\"The value 956 is not of type (UNSIGNED-BYTE 8).\\\\\\\" \\\\\\\" [Condition of type TYPE-ERROR]\\\\ \\\" nil nil) ((\\\\\\\"ABORT\\\\\\\" \\\\\\\"Abort handling SLIME request.\\\\\\\") (\\\\\\\"ABORT\\\\\\\" \\\\\\\"Reduce debugger level (leaving debugger, returning to toplevel).\\\\\\\") (\\\\\\\"TOPLEVEL\\\ \\\\" \\\\\\\"Restart at toplevel READ/EVAL/PRINT loop.\\\\\\\")) ((0 \\ \\\\\"(SB-IMPL::OUTPUT-BYTES/LATIN-1 5 #<FILE-STREAM for \\\\\\\\\\\\\\ \"a constant string\\\\\\\\\\\\\\\" {9833569}> \\\\\\\\\\\\\\\"new debugger invoked on a TYPE-ERROR in thread 14342: The value 956 is not of type (UNSIGNED-BYTE 8). You can type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL. restarts (invokable by number or by possibly-abbreviated name): 0: [ABORT ] Abort handling SLIME request. 1: Reduce debugger level (leaving debugger, returning to toplevel). 2: [TOPLEVEL] Restart at toplevel READ/EVAL/PRINT loop. At this point, SLIME gets pretty confused. Any attempt at making any forward progress (invoking any restarts, etc.) just results in more problems (lands you right back in the debugger). At some point, Emacs seems to get pretty confused, too. I tried this on SBCL without using SLIME (repl in a term window) and it seems to work fine. I'm guessing that there is some sort of problem with Swank communicating with Emacs as it tries to relay the FORMAT output, probably caused by something in the new SBCL Unicode support. Note that you have to use the CBOE website (www.cboe.com) to trigger this. There is something on that page (as simple as a character with the high order bit set??) that triggers this. I have tried other sites and they seem to work fine. I also replaced the (FORMAT T "~A~%" LINE) with (FORMAT T "foo~%") and it prints "foo" very nicely for each line on the web page and does not trigger the bug. If there is a better place/method to file a bug report, let me know. -- Dave Roberts <ldave(a)droberts.com>

3 3

[slime-devel] Loading slime for clisp in xemacs on Windows
by Andrei Stebakov 23 Dec '04

23 Dec '04

Hi, I have xemacs and clisp that came with the cygwin distribution. I just downloaded slime-1.0 and put the following in my custom.el file: (add-to-list 'load-path "C:\\Linux\\slime-1.0\\") (add-hook 'lisp-mode-hook (lambda () (slime-mode t))) (add-hook 'inferior-lisp-mode-hook (lambda () (inferior-slime-mode t))) (setq inferior-lisp-program "clisp -K full") (require 'slime) (slime-setup) Now, when I load slime (via M-x slime) I got: [1]> *** - PARSE-NAMESTRING: syntax error in filename "c:\\Linux\\slime-1.0\\swank-loader.lisp" at position 2 Break 1 [2]> *** - READ from #<IO TERMINAL-STREAM>: there is no package with name "SWANK" Break 2 [3]> "C:\\DOCUME~1\\stebakov\\LOCALS~1\\Temp\\slime.2036" Break 2 [3]> Could anyone give me some info what I can do to fix it? Thanks, Andrei

5 9

[slime-devel] Daily ChangeLog diff
by lgorrie＠common-lisp.net 21 Dec '04

21 Dec '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.593 slime/ChangeLog:1.594 --- slime/ChangeLog:1.593 Thu Dec 16 23:25:23 2004 +++ slime/ChangeLog Tue Dec 21 14:48:40 2004 @@ -1,3 +1,8 @@ +2004-12-16 Martin Simmons <martin(a)xanalys.com> + + * swank-lispworks.lisp (create-socket): Work around bug in + comm::create-tcp-socket-for-service on Mac OS LW 4.3. + 2004-12-16 Edi Weitz <edi(a)agharta.de> * slime.el (slime-complete-symbol*): Bind

1 0

Re: [slime-devel] Modification to swank-lispworks.lisp for LW Personal on Mac OS X
by Bill_Clementson＠peoplesoft.com 21 Dec '04

21 Dec '04

martin(a)xanalys.com writes: > >>>>> On Sun, 19 Dec 2004 20:26:05 -0700, Bill_Clementson(a)peoplesoft.com said: > > Bill> I've just posted about setting up Emacs/SLIME with LW Personal on Mac OS X > Bill> on my blog. There was one minor change needed to "swank-lispworks.lisp" in > Bill> order to get it to work with LW Personal on Mac OS X. Here's the change > Bill> (sorry it's not a patch but I'm on a different machine at the moment): > > Bill> (defimplementation create-socket (host port) > Bill> (multiple-value-bind (socket where errno) > Bill> #+(and (not lispworks4.1) (not lispworks-personal-edition) (not > Bill> macosx)) > Bill> (comm::create-tcp-socket-for-service port :address host) > Bill> #+lispworks4.1(comm::create-tcp-socket-for-service port) > Bill> #+(and macosx > Bill> lispworks-personal-edition)(comm::create-tcp-socket-for-service port) > > Hi there, > > Is this to fix the "Can't assign requested address" error? That was caused by > a bug in LWM 4.3, so I think the following would be better: > > (defimplementation create-socket (host port) > (multiple-value-bind (socket where errno) > #-(or lispworks4.1 (and macosx lispworks4.3)) > (comm::create-tcp-socket-for-service port :address host) > #+(or lispworks4.1 (and macosx lispworks4.3)) > (comm::create-tcp-socket-for-service port) That works for me! Could one of the SLIME maintainers check Martin's changes into CVS please. Thanks, Bill

2 1

[slime-devel] Modification to swank-lispworks.lisp for LW Personal on Mac OS X
by Bill_Clementson＠peoplesoft.com 20 Dec '04

20 Dec '04

Hi all, I've just posted about setting up Emacs/SLIME with LW Personal on Mac OS X on my blog. There was one minor change needed to "swank-lispworks.lisp" in order to get it to work with LW Personal on Mac OS X. Here's the change (sorry it's not a patch but I'm on a different machine at the moment): (defimplementation create-socket (host port) (multiple-value-bind (socket where errno) #+(and (not lispworks4.1) (not lispworks-personal-edition) (not macosx)) (comm::create-tcp-socket-for-service port :address host) #+lispworks4.1(comm::create-tcp-socket-for-service port) #+(and macosx lispworks-personal-edition)(comm::create-tcp-socket-for-service port) - Bill Clementson

2 1

[slime-devel] cl-indent improvement
by Chris Capel 17 Dec '04

17 Dec '04

I take it cl-indent in the SLIME repository is being maintained separately from Emacs' copy, then? Well, two things. First, the CVS entry for 1.2 says lisp-prefix-match-indentation was changed to default to nil. That's not what happened. It's still t in CVS. Second, one of the problems with cl-indent is that it's not possible to specify in a custom indent method that any random list will be indented like so: (a b c d e f g h i) So, I added an option called &rest-offset that makes proceeding lines indented by the following offset integer. So, if you have a form you want indented like this: (deffoo blah blah (blah blah blah blah blah blah)) all you have to do is (put 'deffoo 'common-lisp-indent-function '(6 4 (&whole 2 &rest-offset 1))). Not bad, eh? On a different note, where should forms like that go? They're part of the macro, conceptually, but part of the editor, technically. If you put it in the source file in a comment, it never gets processed when the system is loaded, but if you put it in your .emacs, no one else benefits from it. Maybe there should be a :SWANK in *features*, so you could do: #+swank (swank:eval-in-emacs '(put 'deffoo ...)) This would be evaluated when the file's loaded, but would it happen when the compiled file is loaded? Chris Capel RCS file: /project/slime/cvsroot/slime/cl-indent.el,v retrieving revision 1.2 diff -r1.2 cl-indent.el 42,48d41 < ;; &rest 1 in lambda-lists doesn't work < ;; -- really want (foo bar < ;; baz) < ;; not (foo bar < ;; baz) < ;; Need something better than &rest for such cases < 344a338,339 > (and (eq tem '&rest-offset) > (integerp (cadr method))) 372a368,369 > ((eq tem '&rest-offset) > (throw 'exit (+ sexp-column (cadr method))))

1 0

[slime-devel] macro indentation
by Chris Capel 17 Dec '04

17 Dec '04

Consider (defmacro deffoo (name bar baz &rest options) ...) With this definition (or any definition using rest, really), this is the indentation you get: (deffoo foo bar baz opt1 opt2) Shouldn't &rest be treated like &body as far as macroindentation is concerned? Chris Capel -- "What is it like to be a bat? What is it like to bat a bee? What is it like to be a bee being batted? What is it like to be a batted bee?" -- The Mind's I (Hofstadter, Dennet)

3 3