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

[slime-devel] can't restart lisp/slime in the same emacs session
by Chisheng Huang 22 Jul '04

22 Jul '04

After doing a cvs update today (2004-07-21 PDT), I'm having trouble restarting Lisp/Slime: 1. "M-x slime" to get a *slime-repl[1]* buffer 2. Type (quite) in *slime-repl[1]* buffer and this message shows up in the mini buffer: error in process sentinel: Connection closed 3. Doing "M-x slime" again will not get a *slime-repl[1]* buffer and the following message shows up in the mini buffer: No Lisp subprocess: see variable `inferior-lisp-buffer'. I hope I do not have to mess with inferior-lisp-buffer everytime I restart Lisp/Slime. I'm using CMUCL-19a-pre3 (x86, Linux) and GNU Emacs 21.2.1. Best, -cph

2 2

[slime-devel] Daily ChangeLog diff
by Luke Gorrie 22 Jul '04

22 Jul '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.481 slime/ChangeLog:1.483 --- slime/ChangeLog:1.481 Mon Jul 19 17:53:57 2004 +++ slime/ChangeLog Wed Jul 21 12:56:54 2004 @@ -1,3 +1,12 @@ +2004-07-21 Luke Gorrie <luke(a)bluetail.com> + + * slime.el (slime-sync-package-and-default-directory): Sync + `default-directory' in the REPL buffer too. + (slime-set-state): Convenience function for setting a connection's + state-name and updating the modeline if appropriate. This function + is called in the right places. + (slime-to-lisp-filename): Use `expand-file-name'. + 2004-07-20 Luke Gorrie <luke(a)bluetail.com> * slime.el (slime-repl-update-banner): Restore old behaviour of @@ -86,6 +95,9 @@ * slime.el (package-updating): Fixing this and other tests I broke with my change to how emacs keeps track of the package prompt string. + (arglist): Fix an test failure under Allegro due to a slight + difference in the way EXCL:ARGLIST returns arglist (no default + values of &optional parameters) 2004-07-16 Luke Gorrie <luke(a)bluetail.com>

1 0

[slime-devel] SLIME for newbies, or should the REPL and SLDB have a menu too?
by Dirk Gerrits 21 Jul '04

21 Jul '04

When SLIME is loaded, and you're editing a Lisp file, the menubar shows a SLIME menu where you can go for the less often used features or forgotten keybindings. The REPL and SLDB, however, lack such menus. If it's not too much trouble, I think these should be added. It may make things (even) more accessible to beginning SLIMErs. (And people like me who forget the keybindings all the time until they become firmly engraved in muscle memory when they have been used a bazillion times. ;)) Regards, Dirk Gerrits

2 1

[slime-devel] Really slow to interrupt in Allegro 6.2
by Peter Seibel 21 Jul '04

21 Jul '04

So when I use C-c C-c (slime-interrupt) to try to interrupt a long-running computation in SLIME it sometimes takes a really long time (order 10-60 seconds) before the debugger pops up. I've also noticed that if I do something that generates an error that will contain a lot of output (such as saying (boundp *y*) instead of (boundp '*y*) when *y* is list of several thousand standard-objects) then it also takes a really long time for the debugger to come up. I'm running Allegro 6.2 on GNU/Linux. Haven't had a chance to try this out in other Lisps but thought I'd throw it out there in case anyone has seen a similar thing or has any ideas how to speed things up. -Peter -- Peter Seibel peter(a)javamonkey.com Lisp is the red pill. -- John Fraser, comp.lang.lisp

2 3

[slime-devel] Slime+cmucl+McClim problem
by Thibault Langlois 20 Jul '04

20 Jul '04

Hello, I could not find a solution for the following problem. When I start a McClim application from slime REPL it uses 100% of cpu. I have done what is suggested in the McCLIM instalation tips. I tried to put (mp::startup-idle-and-top-level-loops) in my init.lisp file, i tried to use (setq slime-multiprocessing t) in my .emacs without success. I think the problem is on the slime side. When I start lisp in a xterm there is no problem. Finally I tried to make a cmucl image with the :init-function argument set to mp::startup-idle-and-top-level-loops. Agains it works fine when used from a xterm but does not when used from slime. I use latest (yesterday) cvs slime, latest McCLIM and cmucl 19a (pre) on linux FC2. Any suggestion ? Thanks. Thibault Langlois

2 2

[slime-devel] Daily ChangeLog diff
by Luke Gorrie 20 Jul '04

20 Jul '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.478 slime/ChangeLog:1.481 --- slime/ChangeLog:1.478 Fri Jul 16 19:28:31 2004 +++ slime/ChangeLog Mon Jul 19 17:53:57 2004 @@ -1,3 +1,49 @@ +2004-07-20 Luke Gorrie <luke(a)bluetail.com> + + * slime.el (slime-repl-update-banner): Restore old behaviour of + using an asynchronous evaluation to setup the REPL. This works + around a problem I'd reintroduced where the first REPL command + uses the wrong keymap. + +2004-07-20 Andreas Fuchs <asf(a)boinkor.net> + + * swank-sbcl.lisp (call-with-compilation-hooks): Trap and report + errors that cause compilation to fail, e.g. read errors. + +2004-07-19 Luke Gorrie <luke(a)bluetail.com> + + * HACKING: Updated. Some notes about Emacs features. + + * slime.el: More major refactoring. + Restructured and documented the networking and protocol code. + (slime-rex-continuations): Now connection-local. + +2004-07-18 Luke Gorrie <luke(a)bluetail.com> + + * slime.el: Major refactoring. + Mostly resectioning and reordering definitions to try and improve + readability. + (slime-get-temp-buffer-create): New utility function to popup a + temporary buffer that automatically has a binding on `q' to + intelligently restore window configuration. Handy, but currently + not applicable to all of our temporary buffers. + (slime-with-chosen-connection): Removed this macro. Consequently + the compilation commands no longer prompt for which connection to + use when given a prefix argument. `slime-switch-to-output-buffer' + still works like that, but for other cases I think the + connection-list buffer is sufficient. + (slime-eval-async): New arglist: (form &optional cont pkg). If the + continuation is unspecified then the evaluation result is ignored, + and if the package is unspecified then (slime-buffer-package) is + used. + (slime-eval): Package arg now defaults to (slime-buffer-package). + (slime-current-package): New name for (slime-buffer-package). No + more caching: returns the buffer-local `slime-buffer-package' if + set, otherwise searches for an `in-package' form. + A consequence of non-caching is that the package name doesn't + appear in the modeline anymore. The simplification is worthwhile + in my opinion. + 2004-07-17 Luke Gorrie <luke(a)bluetail.com> * slime.el (slime-autodoc): If there is a global variable name at @@ -6,6 +52,10 @@ supported. (slime-background-message): Truncate messages to fit on a single echo area line. + (slime-repl-update-banner-p, slime-dont-prompt) + (slime-swank-connection-retries): Removed these unused or unuseful + configuration variables. + Rearranged organised "customize" groups. * swank.lisp (variable-desc-for-echo-area): New function. (arglist-for-echo-area): Return nil if symbol can't be found.

1 0

[slime-devel] make Slime recognize file-compile failures on SBCL
by Andreas Fuchs 20 Jul '04

20 Jul '04

Hi, currently, using slime with SBCL isn't very satisfying when you're fond of the C-c C-k function: if you compile the file and it contains read errors, SLIME will think the file compiled ok and try to load it. This fix requires a recent SBCL (as of 0.8.12.39), but it does this: When the file-compiler aborts, it catches the error, and presents it in the *compiler notes* (or whatever it's called today) buffer; it will not try to load the resulting FASL file. The SBCL file compiler aborts as soon as it hits a reader error; so the note output is not really correct. SLIME users should then feel encouraged to fix the reader error(s) and re-try compilation. Does that make sense? Then this patch is for you: Index: README =================================================================== RCS file: /project/slime/cvsroot/slime/README,v retrieving revision 1.10 diff -u -r1.10 README --- README 16 Jul 2004 23:44:36 -0000 1.10 +++ README 19 Jul 2004 21:26:36 -0000 @@ -12,7 +12,7 @@ In Emacs Lisp: - (add-to-list 'load-path "/the/path/to/this/directory") + (add-to-list 'load-path "/home/asf/dl/cvs/slime") (require 'slime) (setup-slime) Index: swank-sbcl.lisp =================================================================== RCS file: /project/slime/cvsroot/slime/swank-sbcl.lisp,v retrieving revision 1.95 diff -u -r1.95 swank-sbcl.lisp --- swank-sbcl.lisp 13 Jul 2004 18:14:03 -0000 1.95 +++ swank-sbcl.lisp 19 Jul 2004 21:26:37 -0000 @@ -168,6 +168,11 @@ (setq *previous-compiler-condition* condition) (signal-compiler-condition condition context)))) +(defun handle-file-compiler-termination (condition) + "Handle a condition that caused the file compiler to terminate." + (handle-notification-condition + (sb-int:encapsulated-condition condition))) + (defun signal-compiler-condition (condition context) (signal (make-condition 'compiler-condition @@ -176,7 +181,8 @@ (sb-c:compiler-error :error) (sb-ext:compiler-note :note) (style-warning :style-warning) - (warning :warning)) + (warning :warning) + (error :error)) :short-message (brief-compiler-message-for-emacs condition) :references ;; FIXME: delete the reader conditionaloid after sbcl @@ -278,7 +284,8 @@ (defimplementation call-with-compilation-hooks (function) (declare (type function function)) - (handler-bind ((sb-c:compiler-error #'handle-notification-condition) + (handler-bind ((sb-c:fatal-compiler-error #'handle-file-compiler-termination) + (sb-c:compiler-error #'handle-notification-condition) (sb-ext:compiler-note #'handle-notification-condition) (style-warning #'handle-notification-condition) (warning #'handle-notification-condition)) @@ -287,13 +294,12 @@ (defvar *trap-load-time-warnings* nil) (defimplementation swank-compile-file (filename load-p) - (flet ((loadit (fasl-file) (when (and load-p fasl-file) (load fasl-file)))) - (cond (*trap-load-time-warnings* - (with-compilation-hooks () - (loadit (compile-file filename)))) - (t - (loadit (with-compilation-hooks () - (compile-file filename))))))) + (handler-case + (let ((output-file (with-compilation-hooks () + (compile-file filename)))) + (when (and load-p output-file) + (load output-file))) + (sb-c:fatal-compiler-error () nil))) (defimplementation swank-compile-string (string &key buffer position) (let ((form (read-from-string (format nil "(~S () ~A)" 'lambda string)))) Have fun, -- Andreas Fuchs, <asf(a)boinkor.net>, asf(a)jabber.at, antifuchs

2 2

[slime-devel] marco's highly opiniated guide to editing lisp code in emacs
by Marco Baringer 19 Jul '04

19 Jul '04

[it would be nice if someone would point out the "proper" place for this info. i was thinking somewhere on the emacswiki, but i couldn't figure out where.] [i suck at speling] emacs is a very large program, i've been using it as a mail reader, shell, note pad and ide for the past 6 years and i'm still learning new things. for those of you who've never really sat down and looked at emacs' lisp editing functions, here's a guide to what i do. 0a - do not program with western european keyboards layouts, accented chars are used very rarely in code, while puncuation chars are far more common. 0b - learn to touch type. 0c - don't spend 1500 EUR (or more) for a computer and then skimp on the keyboard. 1 - in just about every programming language i've ever used (C, Java, perl, bash and lisp), and in normal prose, paranthesis occur for more often that the square bracket chars. so the first thing to do is to make typing the parenthesis easy: (keyboard-translate ?\( ?\[) (keyboard-translate ?\[ ?\() (keyboard-translate ?\) ?\]) (keyboard-translate ?\] ?\)) This assumes that the [ and ] are easy to reach unshifted chars. 2 - learn to use structured editing. what the hell is structured editing? it's just a way of working which attempts, as much as is usefull, to keep the source code consistently valid. this means, among other things, that you inserting an open parenthesis should insert the closing one as well (idem for "). here's my setup: (defun se:move-past-close-and-just-one-space () "Move past next `)' and ensure just one space after it. If called repeatedly removes any leading white space and moves up the enclosing list. Copyright (c) 2002--2003 Hannu Koivisto <azure(a)iki.fi>. All rights reserved. License: GPL." (interactive) (when (eq 'se:move-past-close-and-just-one-space last-command) (delete-horizontal-space t)) (up-list) (just-one-space) (setq this-command 'se:move-past-close-and-just-one-space)) (define-key slime-mode-map [(?\()] 'insert-parentheses) (define-key slime-mode-map [(?\))] 'se:move-past-close-and-just-one-space) (define-key slime-mode-map [(return)] 'newline-and-indent) 2b - so how do you insert a simple, stand alone, ( or ) char? i happen to have the = char directly below ) and the \ char just to the left of that, so here's what i use: (define-key slime-mode-map [(control ?\=)] (lambda () (interactive) (insert "("))) (define-key slime-mode-map [(control ?\\)] (lambda () (interactive) (insert ")"))) chose something that makes sense to use and which requires at most one key press (though it will probably be modified). even with structured editing you'll still need to insert lone parenthesis every now and then. 2c - if you don't want to be so radical and change the meaning of ( and ) know that, in the default lisp mode, insert-parentheses is bound to M-( and M-) is bound to move-past-close-and-reindent. 3 - when editing lisp code you generally don't work on a line or character basis, but you work by moving around lists. usefull emacs functions: transpose-sexps - this simply swaps the sexp before point with the sexp after point. backward-sexp - mave to before the preceding sexp forward-sexp - move after the following sexp kill-sexp - kill the following sexp backward-up-list - move point to the start of the enclosing list. up-list - move forward out of the current enclosing list down-list - move the just after the open paren of the next list all of these functions take prefix args which specify how many sexps to move around. backward-up-list can take a negative arg which maves out of the enclosing parens but forwards, basically simulating up-list. when editing lisp code you will use these the forward/backward/transpose-sexp functions far more often the the char versions, i'd suggest swapping them so that C-M-t (transpose-sexp) is C-t and C-t (transpose-char) is C-M-t: (define-key slime-mode-map (kbd "C-t") 'transpose-sexps) (define-key slime-mode-map (kbd "C-M-t") 'transpose-chars) (define-key slime-mode-map (kbd "C-b") 'backward-sexp) (define-key slime-mode-map (kbd "C-M-b") 'backward-char) (define-key slime-mode-map (kbd "C-f") 'forward-sexp) (define-key slime-mode-map (kbd "C-M-f") 'forward-char) on a dvorak keyboard the htn keys are located directly under the right hand, so i've gone ahead and put the forward functions on n, on the transpose ones on t and the backward ones on h. 4 - i often want to "wrap" a sexp with more code, when for example, i want to add a let binding to a part of code. fortunetly the insert-parentheses function when given a prefix argument will insert the open paren at point and the clasing paren after the following n sexps. so if we have the point before "(foo) (bar)" then simply typing M-2 ( will give us "((foo) (bar))" with the point directly after the first (newly inserted) paren. 5 - i often also decide to unwrap same part of code (if for example i've decided to remove a let and it's bindings around a particular form. (defun mb:unwrap-next-sexp () "Convert (x ...) to ..." (interactive) (forward-sexp) (backward-delete-char 1) (backward-up-list) (delete-char 1) (kill-sexp) (lisp-indent-line)) 6 - I am lucky enough to spend much of my time programming lisp, so i find these changes worth the effort it took to figure them out and get used to them. YMMV. -- -Marco Ring the bells that still can ring. Forget your perfect offering. There is a crack in everything. That's how the light gets in. -Leonard Cohen

4 4

[slime-devel] Compilation errors and warnings in Allegro?
by Peter Seibel 19 Jul '04

19 Jul '04

So it seems that when I compile in Allegro I don't get the fancy outline view of errors and warnings that I get in SBCL. Is there a deep reason for that or just a SMOP? If the latter, roughly what needs to be done? -Peter -- Peter Seibel peter(a)javamonkey.com Lisp is the red pill. -- John Fraser, comp.lang.lisp

2 1

[slime-devel] A command for M-q
by Thomas F. Burdick 18 Jul '04

18 Jul '04

I missed being able to do M-q to reindent the current defun, like in ILISP. Here's the version I'm using. The patch is against a month-old cvs snapshot, sorry. --- slime.el.~1.258.~ Thu Apr 8 16:41:04 2004 +++ slime.el Sun May 2 17:02:35 2004 @@ -449,6 +449,7 @@ ("\M-." slime-edit-definition :inferior t :sldb t) ("\M-," slime-pop-find-definition-stack :inferior t :sldb t) ("\C-q" slime-close-parens-at-point :prefixed t :inferior t) + ("\M-q" slime-reindent-defun :inferior t) ;; Evaluating ("\C-x\C-e" slime-eval-last-expression :inferior t) ("\C-x\M-e" slime-eval-last-expression-display-output :inferior t) @@ -5523,6 +5524,29 @@ (get symbol 'slime-indent)) (put symbol 'slime-indent indent) (put symbol 'common-lisp-indent-function indent))))) + +(defun slime-reindent-defun (&optional force-text-fill) + "Reindent the current defun, or refill the current paragraph. +If point is inside a comment block, the text around point will be +treated as a paragraph and will be filled with `fill-paragraph'. +Otherwise, it will be treated as Lisp code, and the current defun +will be reindented. If the current defun has unbalanced parens, +an attempt will be made to fix it before reindenting. + +When given a prefix argument, the text around point will always +be treated as a paragraph. This is useful for filling docstrings." + (interactive "P") + (save-excursion + (if (or force-text-fill (slime-beginning-of-comment)) + (fill-paragraph nil) + (let ((start (progn (beginning-of-defun) (point))) + (end (ignore-errors (end-of-defun) (point)))) + (unless end + (forward-paragraph) + (slime-close-all-sexp) + (end-of-defun) + (setf end (point))) + (indent-region start end))))) ;;; Test suite

2 3