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

[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

[slime-devel] CVS sbcl and CVS slime trigger evaluation error
by Zach Beane 18 Jul '04

18 Jul '04

I got the following bits after doing C-c C-c on a defpackage form with today's slime checkout; I backed things up with -D, and 2004-07-14 is the first day the problem occurs. Zach --- *slime-repl[0]* --- ; SLIME 2004-07-13 debugger invoked on a SIMPLE-ERROR in thread 7626: There is no applicable method for the generic function #<STANDARD-GENERIC-FUNCTION SWANK-BACKEND:CALL-WITH-DEBUGGING-ENVIRONMENT (0)> when called with arguments (#). 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 SLIME compilation. 1: Abort handling SLIME request. 2: Reduce debugger level (leaving debugger, returning to toplevel). 3: [TOPLEVEL] Restart at toplevel READ/EVAL/PRINT loop. ((NO-APPLICABLE-METHOD (T)) #<unavailable argument> #<unavailable argument> #<STANDARD-GENERIC-FUNCTION SWANK-BACKEND:SWANK-COMPILE-STRING (0)> ("(defpackage :cl-truetype-system (:use :cl :asdf)) " :BUFFER "cl-truetype.asd" :POSITION 10)) 0] --- *slime-repl[0]* --- --- *slime-events* --- (:emacs-rex (swank:connection-info) nil t 1) (:open-dedicated-output-stream 45295) (:use-sigint-for-interrupt) (:indentation-update (("descend-into" . 1) ("number-dispatch" . 1) ("do-debug-fun-blocks" . 1) ("do-debug-block-locations" . 1) ("do-debug-fun-vars" . 1) ("bind-simple-lexical-method-macros" . 1) ("simple-lexical-method-functions" . 1) ("pv-binding1" . 1) ("bind-lexical-method-functions" . 1) ("dotimes-fixnum" . 1) ("dolist-carefully" . 1) ("fast-lexical-method-functions" . 1) ("modify-cache" . 1) ("doplist" . 2) ("bind-args" . 1) ("pv-binding" . 1) ("dfun-miss" . 1) ("bind-fast-lexical-method-macros" . 1) ("without-interrupts" . 0) ("without-scheduling" . 1) ...)) (:return (:ok (7626 "SBCL" "sbcl" (:asdf :ansi-cl :common-lisp :sbcl :unix :sb-doc :sb-test :sb-thread :ieee-floating-point :x86 :linux :gencgc :stack-grows-downward-not-upward :c-stack-is-control-stack))) 1) (:emacs-rex (swank:compile-string-for-emacs "(defpackage :cl-truetype-system\n (:use :cl :asdf))\n" "cl-truetype.asd" 10) ":cl-truetype-system" t 2) (:read-string 0 1) --- *slime-events* ---

1 1

[slime-devel] First Impressions
by iain.little＠gmail.com 17 Jul '04

17 Jul '04

I finally got around to trying SLIME, by downloading the tarfile at: http://www.common-lisp.net/project/slime/slime-1.0alpha.tar.gz I must say, I am very impressed. It looks like Slime aims to fix almost all of my minor annoyances when using cmucl/ilisp. In particular, I like the way that it opens a different buffer to deal with errors/breaks/etc, rather than dumping a screen full of stuff into the repl buffer. But a few things that I noticed: I like to put testing sexprs in the files that I am working on (I put them in all in a #| |# commented region so that I can evaluate them when I want, but they don't evaluate when I compile/load the file). In ilisp I evaluate them with C-z e, and I gather that the slime equivalent is C-x C-e. The problem is, slime puts the output in the minibuffer. (Ok, ilisp can be set up to do this as well, but I have it disabled.) Often I am printing out a representation of a large enough data structure that the output doesn't even fit in the available space. Also, the way that it disappears as soon as a key is pressed prevents me from refering to it later after I have checked/tested something else. The obvious workaround is to copy/paste the sexpr into the repl buffer, but I do this so often (at least on my current project) that this would be a serious pain. For me, the ideal way for this to work would be for there to be a variable that controls whether or not evaluation output goes to the minibuffer or the repl (Or is there one already? I put a bit of effort into looking and couldn't find anything...). A different evaluation command (maybe: slime-eval-last-expression-in-repl) would also work. I also noticed a couple of indentation issues: (ok, I admit it; I went out of my way to see how slime would indent different things ;-) ) CL> (defmacro test (a &body b) (list* '+ a b)) TEST CL> (test 1 2) 3 Which - as advertised - actually does the right thing (ilisp most certainly doesn't). But, CL> (with-accessors ((a b)) c ...) and CL> (defgeneric test2 (a) (:method-combination progn)) #<Standard-Generic-Function TEST2 (0) {48B3C389}> CL> (defmethod test2 progn ((object a)) ...) still arn't right. (But no loss there; ilisp does exactly the same thing...) The only other thing is that the &body indentation doesn't work in the lisp buffers created by opening a file. But I suspect that is because I still have ilisp installed, and it is clobbering whatever slime does. (Is there a workaround?) Anyway, I hope this helps. Slime has impressed me enough that I'm quite certain that I'll switch to it sometime in the future; the only thing stopping me at the moment is the minibuffer issue. (If nothing else, I'll come up with a fix myself when I have more time; my honours project is all-consuming at the moment.) Regards, Iain

4 3

[slime-devel] Hyperspec lookups with w3m in Slime
by d95-bli＠nada.kth.se 17 Jul '04

17 Jul '04

Not directly related to Slime, but hopefully someone can help me still... I get Hyperspec lookup inside of Slime, by setting browse-url-browser-function to w3m-browse-url. Now, when I do C-c C-d h I get Hyperspec lookup, and with C-c C-d C-d I get a description. My roblem is that the hyperspec lookup insists n popping up in the same window where I invoked it, which conveniently hides whatever I was working on. I would much prefer if the behaviour was closer to the one for C-c C-d C-d, where the description pops up in another window, and an effort is made upon exit to restore the windows to what they were before. I suspect that writing my own browser function is the right way to do it, but I am not an elisp wizard, and am not clear on how to go about it. Any ideas? (Sorry if this is too off-topic for the list.) Björn

5 6

[slime-devel] Daily ChangeLog diff
by Luke Gorrie 17 Jul '04

17 Jul '04

Index: slime/ChangeLog diff -u slime/ChangeLog:1.470 slime/ChangeLog:1.478 --- slime/ChangeLog:1.470 Wed Jul 14 22:02:32 2004 +++ slime/ChangeLog Fri Jul 16 19:28:31 2004 @@ -1,3 +1,61 @@ +2004-07-17 Luke Gorrie <luke(a)bluetail.com> + + * slime.el (slime-autodoc): If there is a global variable name at + point then show its value. + (slime-autodoc-cache-type): Cache type 'full is no longer + supported. + (slime-background-message): Truncate messages to fit on a single + echo area line. + + * swank.lisp (variable-desc-for-echo-area): New function. + (arglist-for-echo-area): Return nil if symbol can't be found. + (close-connection): Close connection before printing error + message. This avoids it getting lost in closed I/O redirection. + + * README, doc/slime.texi: Updated setup instructions. + +2004-07-16 Luke Gorrie <luke(a)bluetail.com> + + * slime.el (slime-conservative-indentation): New variable. When + true (the default) don't auto-learn indentation of def* and with-* + macros. Set to nil if you want to learn them. + (slime-handle-indentation-update): Use it. + + * swank.lisp (known-to-emacs-p): Removed filtering of def* and + with-*. Now handled by Emacs. + + * slime.el (slime-interactive-eval): Changed display of results. + By default the result goes to `slime-message', which leads either + to echo area, temporary buffer, or typeout frame. + With a prefix argument the result is printed to the REPL. + This goes for all commands based on slime-interactive-eval, e.g. + `C-x C-e' and `C-c M-:'. + +2004-07-16 Peter Seibel <peter(a)javamonkey.com> + + * slime.el (package-updating): Fixing this and other tests I broke + with my change to how emacs keeps track of the package prompt + string. + +2004-07-16 Luke Gorrie <luke(a)bluetail.com> + + * swank.lisp (print-connection): print-function for connection + objects. Dumping the indentation-cache was damned ugly with + non-truncated lines (e.g. bug reports on slime-devel). + + * slime.el (slime-setup): New function for installing a + lisp-mode-hook. You can call this from ~/.emacs to setup SLIME. + Takes a `autodoc' keyword argument to enable + slime-autodoc-mode. We can add more keywords in future. + (slime-keys): Moved `slime-reindent-defun' from C-M-q to + C-cM-q. This avoids overriding the standard binding of C-M-q to + index-sexp. + (slime-typeout-frame-properties): Removed some properties: + `name', because it interacts badly with X properties, `left' and + `top' because they don't put the frame anywhere terribly + convenient, and (width . 40) because it makes the frame narrower + than the usual word-wrapping width. + 2004-07-14 Peter Seibel <peter(a)javamonkey.com> * slime.el (slime-lisp-package-prompt-string): Separate SLIME's

1 0

[slime-devel] SLIME automated testing best practices.
by Dirk Gerrits 17 Jul '04

17 Jul '04

Hi. I'm not sure if this is appropriate for this mailing list, but I was wondering if and how people are using SLIME to test their programs. Looking at the SLIME manual, there is nothing that really jumps out at me as a program testing tool. Personally, I tend to have two windows above eachother: the top one displaying an open Lisp file where I edit my definitions, the bottom one displaying the SLIME REPL to test these definitions. The problem with this approach is that I don't save my testcases, ie I have no automated regression testing. How do other SLIMErs deal with this? Regards, Dirk Gerrits

4 12