March 2010 - slime-devel - mailman3.common-lisp.net

[slime-devel] PATCH: Improving interaction with other systems
by Andrey Moskvitin 13 Mar '10

13 Mar '10

Hi, Now Hunchentoot included the following code: (when (find-package :swank) (ignore-errors (eval `(,(find-symbol (string '#:add-hook) :swank) ,(find-symbol (string '#:*connection-closed-hook*) :swank) 'after-close-swank-connection)))) This does not seem reliable, but is absolutely necessary for a more safe debugging remote servers. I propose a patch that will rewrite this code more natural way: #+swank(swank:add-hook swank:*connection-closed-hook* 'after-close-swank-connection) See attachment. Moskvitin Andrey

2 1

[slime-devel] Daily ChangeLog diff
by Helmut Eller 13 Mar '10

13 Mar '10

Index: slime/contrib/ChangeLog diff -u slime/contrib/ChangeLog:1.356 slime/contrib/ChangeLog:1.358 --- slime/contrib/ChangeLog:1.356 Tue Mar 9 18:26:23 2010 +++ slime/contrib/ChangeLog Fri Mar 12 22:08:04 2010 @@ -1,3 +1,13 @@ +2010-03-13 Stas Boukarev <stassats(a)gmail.com> + + * slime-asdf.el: use slime-from-lisp-filename so that slime-tramp can work. + Reported by Peter Stirling. + +2010-03-12 Stas Boukarev <stassats(a)gmail.com> + + * swank-arglists.lisp (extract-local-op-arglists (eql 'labels)): + Fix (labels ((name |))). + 2010-03-09 Stas Boukarev <stassats(a)gmail.com> * swank-arglists.lisp (arglist-ref): Don't error if a &key name isn't

1 0

[slime-devel] slime-asdf
by Peter Stirling 13 Mar '10

13 Mar '10

I read about the new "open-system" repl-command on [1], but found it doesn't support slime-tramp for cross-machine editing (it tries to open the files as if they were on the slime-client machine and not use scp to get the files from the machine with the swank-server). I think the attached patch performs the correct incantation, (it works for me anyway). [1] http://planet.lisp.org/

2 1

[slime-devel] Multiple REPLs
by Sebastian Tennant 12 Mar '10

12 Mar '10

Hi all, In buffer *Slime Connections* the marked entry is the 'default' REPL. What's the significance of 'default' status? Is it that in the case of multiple REPLS on a single lisp instance, it determines which REPL receives the lisp output? And if the above thinking is correct, does it have any significance at all when each REPL is on a separate lisp instance? Regards, Seb -- Emacs' AlsaPlayer - Music Without Jolts Lightweight, full-featured and mindful of your idyllic happiness. http://home.gna.org/eap

3 3

[slime-devel] just testing
by Helmut Eller 12 Mar '10

12 Mar '10

nothing to see

1 0

[slime-devel] Daily ChangeLog diff
by Helmut Eller 12 Mar '10

12 Mar '10

Index: slime/ChangeLog diff -u slime/ChangeLog:1.2033 slime/ChangeLog:1.2035 --- slime/ChangeLog:1.2033 Wed Mar 10 06:30:22 2010 +++ slime/ChangeLog Thu Mar 11 04:05:50 2010 @@ -1,5 +1,20 @@ 2010-03-10 Tobias C. Rittweiler <tcr(a)freebits.de> + * swank.lisp (signal-interrupt): Removed. + (interrupt-worker-thread): Slurp in definition of + signal-interrupt. No need for invoke-or-queue-interrupt twice in + case we do not use threads. Thus micro-prettification of + backtraces. + +2010-03-10 Tobias C. Rittweiler <tcr(a)freebits.de> + + * swank-ecl.lisp (source-location): Move call to + TRANSLATE-LOGICAL-PATHNAME from here into MAKE-FILE-LOCATION + because locations-via-annotations may now also involve logical + pathnames. + +2010-03-10 Tobias C. Rittweiler <tcr(a)freebits.de> + * swank-ecl.lisp (*original-sigint-handler*) (install-sigint-handler): Deleted; we directly implement call-with-user-break-handler instead.

1 0

[slime-devel] Bug Report: SB-KERNEL:SIMPLE-PACKAGE-ERROR not invoking the debugger
by Russ Tyndall 11 Mar '10

11 Mar '10

Slime Version: 3/10 from git mirror of CVS SBCL 1.0.36 GNU Emacs 22.1.1 and 23.x.x (checked on two different versions). Using the latest version of slime, executing the following code in the slime repl results in a totally locked lisp that can't be quit or interacted with via slime. There is also nothing of interest in the backing SBCL terminal. (make-package :short-name) (make-package :long-name :nicknames (list :short-name)) Executing these same forms in the sbcl command line repl correctly results in a debugger asking for me to abort being stupid. The last slime event we see is: (:emacs-rex (swank:listener-eval "(make-package :long-name :nicknames (list :short-name))\n") "COMMON-LISP-USER" :repl-thread 6) This is my setup: (slime-setup '(slime-fancy slime-asdf slime-tramp slime-indentation)) Thanks for the great software, SLIME is a pleasure to use, Russ Tyndall

2 1

[slime-devel] Daily ChangeLog diff
by Helmut Eller 11 Mar '10

11 Mar '10

Index: slime/ChangeLog diff -u slime/ChangeLog:1.2032 slime/ChangeLog:1.2033 --- slime/ChangeLog:1.2032 Tue Mar 9 19:02:53 2010 +++ slime/ChangeLog Wed Mar 10 06:30:22 2010 @@ -1,3 +1,12 @@ +2010-03-10 Tobias C. Rittweiler <tcr(a)freebits.de> + + * swank-ecl.lisp (*original-sigint-handler*) + (install-sigint-handler): Deleted; we directly implement + call-with-user-break-handler instead. + (call-with-user-break-handler): New. Correctly interrupt main + thread instead of newly spawned handle-signal thread on SIGINT. + (make-interrupt-handler): New helper. + 2010-03-09 Stas Boukarev <stassats(a)gmail.com> * swank-ccl.lisp (emacs-inspect function): Print closed over variables @@ -62,7 +71,7 @@ it boils down to SIGNAL-INTERRUPT which uses USE-THREADS-P which needs a connection. (install-fd-handler): Adapted accordingly. - (simple-serve-event): Adapted accordingly. Additionally, remove + (simple-serve-requests): Adapted accordingly. Additionally, remove superfluous WITH-SWANK-PROTOCOL-HANDLER as that's established by WITH-CONNECTION already. (simple-repl): Show "abort inferior lisp" restart only if not a

1 0

[slime-devel] sldb-initial-restart-limit
by Robert Goldman 10 Mar '10

10 Mar '10

The current definition of this custom is as follows: (defcustom sldb-initial-restart-limit 6 "Maximum number of restarts to display initially." :group 'slime-debugger :type 'integer) Reading the way this is used, I believe this could be changed to (defcustom sldb-initial-restart-limit 6 "Maximum number of restarts to display initially. Will always display all restarts if NIL." :group 'slime-debugger :type '(choice integer (const :tag "Show all" nil)) I don't venture to offer this as a patch because I'm not an expert with customize. HtH, r

1 0

[slime-devel] slime-compile-file gets error in ACL 8.2
by Takuo Watanabe 10 Mar '10

10 Mar '10

Dear Slime Developers I am using Slime with the following environment: Slime 2010-03-01 (CVS head) Emacs 22.3.1 (i386-apple-darwin9.8.0, Carbon Version 1.6.0) Allegro CL Professional 8.2 (allegro-8.2a-ics-macosx-x86) Mac OS X (10.6.2), MacBook After upgrading Allegro to 8.2, slime-compile-file (also slime-compile-and-load-file) gets errors like this: ------------------------------------------------------------ `(6 6)' is not of the expected type `NUMBER' [Condition of type TYPE-ERROR] ------------------------------------------------------------ The backtrace of the error looks like: ------------------------------------------------------------ Backtrace: 0: (ERROR TYPE-ERROR :DATUM (6 6) :EXPECTED-TYPE NUMBER ...) 1: (SWANK-BACKEND::LOCATION-FOR-WARNING #<SIMPLE-WARNING @ #x20a331c2>) Locals: CONDITION = #<SIMPLE-WARNING @ #x20a331c2> SWANK-BACKEND::POS = (6 6) NIL = (:FILE "/Users/takuo/foo.lisp") NIL = #<STANDARD-CLASS READER-ERROR> SWANK-BACKEND::LOC = (#P"/Users/takuo/foo.lisp" 6 6) SWANK-BACKEND::FILE = #P"/Users/takuo/foo.lisp" 2: (SWANK-BACKEND::HANDLE-COMPILER-WARNING #<SIMPLE-WARNING @ #x20a331c2>) 3: (SIGNAL #<SIMPLE-WARNING @ #x20a331c2>) ------------------------------------------------------------ The source of this error seems to be the change of the format of condition location (i.e., the value of the :loc property provided with the excl::plist slot in a warning object). For example, in the previous ACL (8.1), the value of :loc property looks like this. ------------------------------------------------------------ CL-USER> (handler-case (compile-file "foo.lisp") (warning (w) (getf (slot-value w 'excl::plist) :loc))) (#P"foo.lisp" . 6) ------------------------------------------------------------ The car part of the value is the file name and the cdr is the warning position in the file. After upgrading to 8.2, the value of the above expression becomes (#P"foo.lisp" 6 6). I don't know about the difference among these two numbers. Anyway, the function location-for-warning in swank-allegro.lisp assumes that the cdr part of the value should be a number. The (kludgey) patch I applied is the following. I'm not sure this is really a solution to the problem. ------------------------------------------------------------ --- swank-allegro.lisp 2010-03-02 12:25:46.000000000 +0900 +++ swank-allegro.lisp.orig 2010-03-02 12:25:24.000000000 +0900 @@ -279,7 +279,7 @@ (destructuring-bind (file . pos) loc (make-location (list :file (namestring (truename file))) - (list :position (1+ (if (listp pos) (car pos) pos)))))) + (list :position (1+ pos))))) (t (make-error-location "No error location available."))))) ------------------------------------------------------------ Sorry, I didn't check any other platforms (Linux, Windows, etc.). I only have a license for the OS X edition. Cheers Takuo Watanabe

3 2