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

[slime-devel] Daily ChangeLog diff
by lgorrie＠common-lisp.net 29 Nov '04

29 Nov '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.581 slime/ChangeLog:1.582 --- slime/ChangeLog:1.581 Thu Nov 25 20:09:54 2004 +++ slime/ChangeLog Mon Nov 29 18:36:37 2004 @@ -1,3 +1,36 @@ +2004-11-29 Lynn Quam <quam(a)ai.sri.com> + + * slime.el (slime-global-variable-name-p): Allow optional + "<package-name>:" or "<package-name>::". + +2004-11-29 Chris Capel <pdf23ds(a)gmail.com> + + * swank.lisp (macro-indentation): Ignore &whole, &aux, and + &environment args. + +2004-11-29 Helmut Eller <heller(a)common-lisp.net> + + * slime.el (slime-repl-wrap-history): New user variable. + (slime-repl-history-replace): Implement wrap around. + (slime-repl-easy-menu): Fix binding for "Next Input". Reported by + Surendra Singhi. + + * swank-lispworks.lisp (list-callers-internal): Return the + function if dspec:object-dspec returns nil. + (xref-results): Previously, functions for which + dspec:dspec-definition-locations returned nil were ignored. + Include them with a unknown source-location. + + * swank-abcl.lisp, swank-allegro.lisp, swank-clisp.lisp, + swank-cmucl.lisp, swank-openmcl.lisp, swank-sbcl.lisp, + swank-lispworks (accept-connection): The :external-format argument + defaults now to :iso-latin-1-unix. + +2004-11-26 Helmut Eller <heller(a)common-lisp.net> + + * swank-cmucl.lisp (read-into-simple-string): Use #-cmu19 instead + of #+cmu18e. + 2004-11-25 Chris Capel <pdf23ds(a)gmail.com> * slime.el (slime-indent-and-complete-symbol): Echo the arglist if @@ -28,14 +61,14 @@ connection. It's almost impossible to run the debugger inside the control-thread, so let it crash instead. A backtrace would be nice, though. - (cleanup-connection-threads): Can know be called in the + (cleanup-connection-threads): Can now be called in the control-thread. Add a check to avoid thread suicide. (start-swank-server-in-thread): Fix the call to start-server. * swank-sbcl.lisp (%thread-state-slot, %thread-state): Refactored from thread-status. (thread-status): Use it. - (all-threads): Exclude return zombies. + (all-threads): Exclude zombies. 2004-11-24 Helmut Eller <heller(a)common-lisp.net>

1 0

[slime-devel] Incompatibility in latest UCW and SLIME
by Chris Capel 29 Nov '04

29 Nov '04

I'm tracking the head of both UCW and SLIME, with SBCL. SLIME recently added the external-format keyword to a bunch of different functions, including swank-backend:accept-connection. Accept-connection is used by UCW as its socket interface. It doesn't pass an encoding. Either UCW should pass an encoding, or accept-connection should pick up on the default *CODING-SYSTEM*, or swank-sbcl.lisp should pick up on it somewhere. The error I'm getting is in swank-sbcl.lisp, make-socket-io-stream. External-formal is nil, and thus falls through the ECASE. Right now, only swank:start-server, create-server, and create-swank-server pick up on the default coding system. The argument could be made that SLIME needs a sensible default in this situation. The argument could be made that UCW needs to be in control of this sort of thing. I'll leave it to you all. 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)

2 1

[slime-devel] Improvement request to M-.
by Chris Capel 29 Nov '04

29 Nov '04

I noticed that it doesn't handle class names. Would it be possible to have it find the definition for a class if it doesn't find another meaning of the word? It also seems to skip over defparameters. Chris Capel

2 4

[slime-devel] problem with C-m-x, eval-region and a host of other functions.
by Sundar Narasimhan 26 Nov '04

26 Nov '04

Hi, All: I updated my copy of slime to the dev-head (perhaps this is an unwise thing to do :), and notice that on fedora core 2 with cmucl, a lot of basic commands such as eval-defun, eval-region etc. have stopped working. In the inferior lisp window, when the connection is lost, I see printouts of the form: ;; Connection to Emacs lost. ;; [ Error in function LISP::ASSERT-ERROR: Short read: length=798 pos=506] My cursory examination of the relevant code has me confused.. since I can't quite see (yet) how this could happen. Is there a way I can debug this or provide you all more info? (perhaps tracing with-transcript or dispatch functions?) Thanks, -Sundar

2 2

[slime-devel] Daily ChangeLog diff
by Luke Gorrie 25 Nov '04

25 Nov '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.579 slime/ChangeLog:1.581 --- slime/ChangeLog:1.579 Wed Nov 24 21:29:39 2004 +++ slime/ChangeLog Thu Nov 25 20:09:54 2004 @@ -1,3 +1,42 @@ +2004-11-25 Chris Capel <pdf23ds(a)gmail.com> + + * slime.el (slime-indent-and-complete-symbol): Echo the arglist if + there's no symbol before point. Don't complete after parens. + (slime-echo-arglist): Factorized from slime-space. + (slime-space): Use it. + (slime-repl-history-replace): Clear the input at the end of the + history. + + * swank.lisp (arglist-to-string): Don't show &whole, &aux and + &environment args. + (clean-arglist): New function. + +2004-11-25 Helmut Eller <heller(a)common-lisp.net> + + * slime.el (slime-net-coding-system): Emacs does funky encoding + for `raw-text-unix' use `binary' instead. + (slime-safe-encoding-p): New function. + (slime-net-send): Use it and don't try to send stuff which can't + be decoded by Lisp. + (slime-inferior-lisp-program-history): XEmacs compatibility: + declare it as a variable. + (slime-xref-mode): In Emacs 21, set delayed-mode-hooks to nil + because we don't want to run the lisp-mode-hook. Reported by + Chris Capel. + + * swank.lisp (dispatch-loop): Catch errors and close the + connection. It's almost impossible to run the debugger inside the + control-thread, so let it crash instead. A backtrace would be + nice, though. + (cleanup-connection-threads): Can know be called in the + control-thread. Add a check to avoid thread suicide. + (start-swank-server-in-thread): Fix the call to start-server. + + * swank-sbcl.lisp (%thread-state-slot, %thread-state): Refactored + from thread-status. + (thread-status): Use it. + (all-threads): Exclude return zombies. + 2004-11-24 Helmut Eller <heller(a)common-lisp.net> * slime.el (slime-start-and-load): Use vanilla comint instead of @@ -11,7 +50,7 @@ inf-lisp. (slime-net-coding-system): Use find-coding-system in XEmacs. coding-system-p means something different here. - (slime-repl-mode-map): XEmacs compatibilty: use (kbd "C-<up>") + (slime-repl-mode-map): XEmacs compatibility: use (kbd "C-<up>") instead of [C-up]. * swank.lisp (inspect-for-emacs-list): subseq on improper lists

1 0

[slime-devel] [PATCH] arglist feature
by Chris Capel 25 Nov '04

25 Nov '04

I added a sort of slime-indent-and-complete-symbol-and-show-arglist function that shows the arglist like slime-space does if there's a whitespace character before the point, and completes the symbol otherwise. Oh, and indents it if it needs to be indented. To accomplish this, I refactored slime-space to use slime-echo-arglist-for-enclosing so that I could use the same function in slime-indent (renamed from slime-indent-and-complete-symbol). Slime-indent takes a couple customization options so you can disable either the symbol completion or arglist showing. A DWIM tab key! Now I don't need slime-space or slime-return! Awesome. Patch attached. Chris Capel

2 3

[slime-devel] XREF troubles
by Chris Capel 25 Nov '04

25 Nov '04

Whever I do a M-. on a method, up pops an XREF window. Nice. But there's a bit of a problem. It's in slime-mode. Slime-mode stomps on XREF's keys. Not nice. And this is a bit odd, because I was looking at the hook for XREF, and it disables slime-mode. But, since forever, I've had a lisp-mode hook that enables slime-mode, and I guess it gets run after the XREF hook. I think I read somewhere in the slime docs that you're supposed to set up your lisp-mode hook this way. So is this a bug, or do I need a different configuration? Chris Capel

2 2

[slime-devel] [PATCH] improvement to arglist
by Chris Capel 25 Nov '04

25 Nov '04

I consider the &whole keyword in a macro lambda list to be an implementation detail, and not something that needs to bother the person calling the macro (or something that needs to interfere with slime's automatic macro indention). I also consider &aux useless for the same purposes. So I changed "(defun arglist-to-string" in swank to remove "&whole var" and the rest of the lambda list after and including &aux from the lambda list. All in all, two extra functions, and a macro used by both; two new test cases for lambda list printing. Chris Capel

2 1

[slime-devel] [PATCH] improvement to repl history
by Chris Capel 25 Nov '04

25 Nov '04

I noticed that when you scroll through the history in the slime REPL, when you reach the end it doesn't clear the commandline, as bash and other shells do. I think this is rather inconvenient, so I patched it up. I made it so there's a blank space at the beginning, too, just because. No reason not to. I added a customization setting under Repl, that defaults to NIL. When it's true, the history will wrap around when the end is reached and you keep going. Chris Capel

2 1

[slime-devel] Fwd: Showing argument lists on return
by Svein Ove Aas 25 Nov '04

25 Nov '04

Ok, let's try that again, with me subscribed... ---------- Forwarded message ---------- From: Svein Ove Aas <sveina(a)gmail.com> Date: Tue, 23 Nov 2004 13:42:19 +0000 Subject: Showing argument lists on return To: slime-devel(a)common-lisp.net Cc: sveina(a)gmail.com Recently I've noticed I often press return-backspace-space in quick succession to keep the argument list display around. I put the following snippet in .emacs to fix it; I don't have time to make a proper patch of it now due to exams, but maybe some of you will find it handy. There are a number of obvious problems with it, and probably a few less obvious ones. Have fun! Also, I'm not subscribed to slime-devel, so please keep me CCed in any replies. ----- (define-key slime-mode-map (kbd "RET") 'slime-return) (defun slime-return () (interactive) (unwind-protect (if (slime-connected-p) (when (and slime-space-information-p (or (not (slime-busy-p)) ;; XXX should we enable this? ;; (not slime-use-sigint-for-interrupt)) )) (let ((names (slime-enclosing-operator-names))) (when names (slime-eval-async `(swank:arglist-for-echo-area (quote ,names)) (lambda (message) (if message (slime-message "%s" message))))))) (slime-message "Connect to a Lisp to see the argument list.")) (newline-and-indent))) (defun slime-repl-return () (interactive) (slime-check-connected) (assert (<= (point) slime-repl-input-end-mark)) (cond ((get-text-property (point) 'slime-repl-old-input) (slime-repl-grab-old-input)) (current-prefix-arg (slime-repl-send-input)) (slime-repl-read-mode ; bad style? (slime-repl-send-input t)) ((slime-input-complete-p slime-repl-input-start-mark slime-repl-input-end-mark) (slime-repl-send-input t)) (t (unwind-protect (when (and slime-space-information-p (slime-connected-p) (or (not (slime-busy-p)) ;; XXX should we enable this? ;; (not slime-use-sigint-for-interrupt)) )) (let ((names (slime-enclosing-operator-names))) (when names (slime-eval-async `(swank:arglist-for-echo-area (quote ,names)) (lambda (message) (if message (slime-message "%s" message))))))) (newline-and-indent)))))

2 1