November 2009 - slime-devel - mailman3.common-lisp.net

[slime-devel] Daily ChangeLog diff
by Helmut Eller 03 Nov '09

03 Nov '09

Index: slime/ChangeLog diff -u slime/ChangeLog:1.1896 slime/ChangeLog:1.1898 --- slime/ChangeLog:1.1896 Wed Oct 28 16:52:20 2009 +++ slime/ChangeLog Fri Oct 30 15:39:34 2009 @@ -1,3 +1,16 @@ +2009-10-30 Stas Boukarev <stassats(a)gmail.com> + + * swank-loader.lisp (setup): Use EXT:PROBE-DIRECTORY on Clisp, + because PROBE-FILE doesn't want to work on directories. + Patch by Dirk Sondermann. + +2009-10-30 Helmut Eller <heller(a)common-lisp.net> + + * swank-openmcl.lisp: Removed. 1.4 is out so no longer needed. + * swank-ccl.lisp: Update accordingly. + + * test-all.sh: Removed. Not used in ages. + 2009-10-28 Helmut Eller <heller(a)common-lisp.net> Simpler modeline code. Index: slime/contrib/ChangeLog diff -u slime/contrib/ChangeLog:1.261 slime/contrib/ChangeLog:1.262 --- slime/contrib/ChangeLog:1.261 Wed Oct 21 15:38:49 2009 +++ slime/contrib/ChangeLog Fri Oct 30 19:06:26 2009 @@ -1,3 +1,13 @@ +2009-10-31 Tobias C. Rittweiler <tcr(a)freebits.de> + + `M-x slime-visit-sbcl-bug' will open a browser to visit SBCL's bug + tracker at Launchpad that describes the bug number at + point (#nnnnnn). + + * slime-sbcl-exts.el (slime-visit-sbcl-bug): New. + (slime-read-sbcl-bug): New. + (slime-sbcl-bug-at-point): New. + 2009-10-21 Stas Boukarev <stassats(a)gmail.com> * slime-asdf.el (slime-browse-system): Open the parent directory of

2 2

[slime-devel] *sysdep-files* as var
by Terje Norderhaug 03 Nov '09

03 Nov '09

I propose *sysdep-files* is changed from a parameter to a var, so it can be bound before loading swank-loader.lisp. I need this for an enhanced swank loader that supports a lisp implementation not covered by the current *sysdep-files* options, but I believe it could be generally useful for those tinkering with swank. diff -u slime-2009-11-03/swank-loader.lisp slime-2009-11-03/swank- loader-patched.lisp --- slime-2009-11-03/swank-loader.lisp 2009-11-01 01:15:02.000000000 -0700 +++ slime-2009-11-03/swank-loader-patched.lisp 2009-11-03 11:12:21.000000000 -0800 @@ -32,7 +32,7 @@ :defaults (or *load-pathname* *default-pathname- defaults*)) "The directory where to look for the source.") -(defparameter *sysdep-files* +(defvar *sysdep-files* #+cmu '(swank-source-path-parser swank-source-file-cache swank- cmucl) #+scl '(swank-source-path-parser swank-source-file-cache swank-scl) #+sbcl '(swank-source-path-parser swank-source-file-cache -- Terje Norderhaug

1 0

[slime-devel] slime.el byte-compile notes (emacs 23.1)
by Derrell Piper 03 Nov '09

03 Nov '09

I'm using a byte-compiled slime.el. Just FYI, here's your current compilation notes for emacs 23.1: Compiling file /Users/ddp/.emacs.d/slime/slime.el at Tue Nov 3 09:50:45 2009 Entering directory `/Users/ddp/.emacs.d/slime/' In slime-complete-maybe-restore-window-configuration: slime.el:3633:20:Warning: `last-command-char' is an obsolete variable (as of Emacs at least 19.34); use `last-command-event' instead. In slime-next-line/not-add-newlines: slime.el:4698:10:Warning: `next-line' used from Lisp code That command is designed for interactive use only In sldb-recenter-region: slime.el:5709:27:Warning: `next-line' used from Lisp code That command is designed for interactive use only In slime-add-local-hook: slime.el:8555:11:Warning: `make-local-hook' is an obsolete function (as of Emacs 21.1); not necessary any more. Compiling no file at Tue Nov 3 09:50:59 2009

1 0

[slime-devel] backtrace --more-- broken
by Derrell Piper 03 Nov '09

03 Nov '09

The backtrace "--more--" tag no longer works. If you "n" off the bottom of a long BT, the "--more--" tag goes away but nothing happens. If you try to "p" back to one of the existing frames, you get this in the minibuffer and the cursor remains on the blank line: slime-property-bounds: Assertion failed: (get-text-property (point) prop) Derrell

2 1

[slime-devel] compile-file-if-needed fails if fasl already exists (Clozure)
by Derrell Piper 03 Nov '09

03 Nov '09

If I try to compile and load a file that's already compiled in Clozure, say doing this twice: (swank:compile-file-if-needed "/Users/ddp/src/lisp/test.lisp" t) On the second time around, swank signals an error in swank:collect- notes because Clozure returns a pathname of the fasl file instead of T. Clozure's behavior seems rational. So since the underlying defstruct specifies :type boolean for successp, here's how I fixed this: fluffy% diff -n swank.lisp.~1.671.~ swank.lisp d2786 2 a2787 2 (let ((successp (if successp t nil))) (make-compilation-result (reverse notes) successp seconds))))) (defun collect-notes (function) (let ((notes '())) (multiple-value-bind (successp seconds) (handler-bind ((compiler-condition (lambda (c) (push (make-compiler-note c) notes)))) (measure-time-interval function)) (let ((successp (if successp t nil))) (make-compilation-result (reverse notes) successp seconds)))))

2 1

[slime-devel] Daily ChangeLog diff
by Helmut Eller 03 Nov '09

03 Nov '09

Index: slime/ChangeLog diff -u slime/ChangeLog:1.1905 slime/ChangeLog:1.1907 --- slime/ChangeLog:1.1905 Mon Nov 2 02:47:02 2009 +++ slime/ChangeLog Mon Nov 2 11:17:48 2009 @@ -1,3 +1,14 @@ +2009-11-02 Tobias C. Rittweiler <tcr(a)freebits.de> + + * slime.el (slime-end-of-symbol): Make sure not to move on #'foo. + ([test] sexp-at-point.1): New test case. + +2009-11-02 Helmut Eller <heller(a)common-lisp.net> + + * swank.lisp (without-interrupts): Removed. No longer used. + * swank-backend.lisp (call-without-interrupts): Removed. + Update backends accoringly. + 2009-11-02 Helmut Eller <heller(a)common-lisp.net> CCL's lap-functions don't have source-notes but the name often Index: slime/contrib/ChangeLog diff -u slime/contrib/ChangeLog:1.267 slime/contrib/ChangeLog:1.269 --- slime/contrib/ChangeLog:1.267 Sat Oct 31 18:41:04 2009 +++ slime/contrib/ChangeLog Mon Nov 2 11:24:45 2009 @@ -1,3 +1,19 @@ +2009-11-02 Tobias C. Rittweiler <tcr(a)freebits.de> + + * slime-parse.el (slime-make-form-spec-from-string): Break out of + the loop if we're at unbalanced parentheses. + (slime-compare-character-syntax): New helper. + (slime-parse-form-upto-point): Use it. + (slime-incomplete-form-at-point): Revert change. + ([test] form-upto-point.1): New test case. + +2009-11-02 Stas Boukarev <stassats(a)gmail.com> + + * slime-parse.el (slime-incomplete-form-at-point): Concatenate " )" + not just ")", because the form's last char may be \, and the parenthesis + will be escaped. + That fixes a bug reported by Ariel Badichi. + 2009-10-31 Tobias C. Rittweiler <tcr(a)freebits.de> * slime-fontifying-fu.el (slime-search-suppressed-forms-internal):

1 0

[slime-devel] Emacs may freeze
by Ariel Badichi 02 Nov '09

02 Nov '09

Hello, With the new commit from 2009-10-31, I can reproducibly cause GNU Emacs 22.2.1 to lock-up. The commit is huge, and I didn't have the time to further localize the problem. To see the problem, start Emacs/SLIME and in the SLIME scratch buffer type the following: (char= #\ Ariel

3 2

[slime-devel] Daily ChangeLog diff
by Helmut Eller 02 Nov '09

02 Nov '09

Index: slime/ChangeLog diff -u slime/ChangeLog:1.1903 slime/ChangeLog:1.1905 --- slime/ChangeLog:1.1903 Sat Oct 31 18:41:03 2009 +++ slime/ChangeLog Mon Nov 2 02:47:02 2009 @@ -1,3 +1,19 @@ +2009-11-02 Helmut Eller <heller(a)common-lisp.net> + + CCL's lap-functions don't have source-notes but the name often + has. E.g. ccl::%fixnum-truncate. Use names as last resort. + + * swank-ccl.lisp (function-name-source-note): New function. + (pc-source-location): Use it. + +2009-11-02 Stas Boukarev <stassats(a)gmail.com> + + * swank.lisp (tokenize-symbol-thoroughly): Return NIL + instead of throwing an error. + (parse-symbol): Handle null result of tokenize-symbol-thoroughly. + + This fixes a bug reported by Derrell Piper. + 2009-10-31 Tobias C. Rittweiler <tcr(a)freebits.de> * slime.el (slime-bug): Deleted.

1 0

[slime-devel] swank:tokenize-symbol-thoroughly
by Derrell Piper 02 Nov '09

02 Nov '09

I'm running cvs head and using slime-fancy (which includes slime- autodoc). swank:tokenize-symbol-thoroughly is looking for a package specifier and complains if it sees more than two colons. It thinks it's parsing a symbol but this chunk of code causes it to signal an error when called with the format control string: (with-slots (n-start n-pairs min-word legal-chars pairs pair-table) db (when verbose (format t "n-start: ~d, n-pairs: ~d, min-word: ~d~%" n-start n-pairs min-word)) (dotimes (i n-pairs) (NB: this obviously isn't a fully formed sexp, just the the smallest text blob I could reproduce this with...) To reproduce this, load slime (w/ autodoc or fancy), bring the above four lines up in a lisp buffer and try to C-n to the last line (backtrace below). There's something about the first two lines that's relevant. If you delete either the (with-slots...) or (when verbose...) (or both) this doesn't happen. More than two colons in "\"n-start: ~d, n-pairs: ~d, min-word: ~d~%\"" [Condition of type SIMPLE-ERROR] Restarts: 0: [ABORT] Return to SLIME's top level. 1: [ABORT-BREAK] Reset this thread 2: [ABORT] Kill this thread Backtrace: 0: (SWANK::TOKENIZE-SYMBOL-THOROUGHLY "\"n-start: ~d, n-pairs: ~d, min-word: ~d~%\"") Locals: STRING = "\"n-start: ~d, n-pairs: ~d, min-word: ~d~%\"" PACKAGE = "\"N-START" SWANK::TOKEN = " ~D, N-PAIRS ~D, MIN-WORD" SWANK::BACKSLASH = NIL SWANK::VERTICAL = NIL SWANK::INTERNP = T CHAR = #\: #:LOOP-ACROSS-VECTOR-2378 = "\"n-start: ~d, n-pairs: ~d, min- word: ~d~%\"" #:LOOP-ACROSS-INDEX-2379 = 36 #:LOOP-ACROSS-LIMIT-2380 = 42 1: (SWANK::PARSE-SYMBOL "\"n-start: ~d, n-pairs: ~d, min-word: ~d~% \"" #<Package "COMMON-LISP-USER">) 2: (SWANK::READ-CONVERSATIVELY "\"n-start: ~d, n-pairs: ~d, min- word: ~d~%\"") 3: (SWANK::PARSE-RAW-FORM ("format" "t" "\"n-start: ~d, n-pairs: ~d, min-word: ~d~%\"" "n-start" "n-pairs" "min-word")) 4: (SWANK::PARSE-RAW-FORM ("when" "verbose" ("format" "t" "\"n- start: ~d, n-pairs: ~d, min-word: ~d~%\"" "n-start" "n-pairs" "min- word"))) 5: (SWANK::PARSE-RAW-FORM ("with-slots" ("n-start" "n-pairs" "min- word" "legal-chars" "pairs" "pair-table") "db" ..))) 6: ((:INTERNAL SWANK:ARGLIST-FOR-ECHO-AREA)) 7: (SWANK::CALL-WITH-BUFFER-SYNTAX NIL #<COMPILED-LEXICAL-CLOSURE (:INTERNAL SWANK:ARGLIST-FOR-ECHO-AREA) #x3000415FF41F>) 8: (CCL::CALL-CHECK-REGS SWANK:ARGLIST-FOR-ECHO-AREA ("with- slots" ("n-start" "n-pairs" "min-word" "legal-chars" "pairs" "pair- table") "db" ..)) :PRINT-RIGHT-MARGIN 1000 :PRINT-LINES 1) 9: (SWANK::EVAL-FOR-EMACS (SWANK:ARGLIST-FOR-ECHO-AREA ..)))) "COMMON-LISP-USER" 9) 10: ((:INTERNAL SWANK::SPAWN-WORKER-THREAD)) 11: (#<Anonymous Function #x30004124EB3F> #<Compiled-function SWANK:SWANK-DEBUGGER-HOOK #x30004138413F> #<Compiled-function (:INTERNAL SWANK::SPAWN-WORKER-THREAD) (Non-Global) #x30004131288F>) 12: (SWANK::CALL-WITH-BINDINGS ((*STANDARD-OUTPUT* . #1=#<SWANK- BACKEND::SLIME-OUTPUT-STREAM #x300041427ABD>) (*STANDARD-INPUT* . #<SWANK-BACKEND::SLIME-INPUT-STREAM #x300041427E7D>) ..))) #<COMPILED- LEXI.. 13: (SWANK::CALL-WITH-CONNECTION #<CONNECTION #x3000413BC66D> #<Compiled-function (:INTERNAL SWANK::SPAWN-WORKER-THREAD) (Non- Global) #x30004131288F>) 14: (SWANK::CALL-WITH-BINDINGS NIL #<COMPILED-LEXICAL-CLOSURE (:INTERNAL SWANK::SPAWN-WORKER-THREAD) #x3000415FFAFF>) 15: (SWANK-BACKEND::ENTRY) 16: (CCL::RUN-PROCESS-INITIAL-FORM #<PROCESS worker(14) [Active] #x300041568DBD> (#<COMPILED-LEXICAL-CLOSURE (:INTERNAL CCL::%PROCESS- RUN-FUNCTION) #x300041568B3F>)) 17: ((:INTERNAL (CCL::%PROCESS-PRESET-INTERNAL (PROCESS))) #<PROCESS worker(14) [Active] #x300041568DBD> (#<COMPILED-LEXICAL-CLOSURE (:INTERNAL CCL::%PROCESS-RUN-FUNCTION) #x300041568B3F>)) 18: ((:INTERNAL CCL::THREAD-MAKE-STARTUP-FUNCTION))

2 1

[slime-devel] slime-selector ?r and slime-repl ordering
by Derrell Piper 01 Nov '09

01 Nov '09

A note on slime-setup ordering. I recently noticed that slime- selector 'r' quit taking me to the REPL and instead started taking me to the *inferior-lisp* buffer. The reason turned out to be the ordering of the slime-setup in my .emacs: This: (slime-setup '(slime-fancy slime-asdf inferior-slime)) ...does not work, whereas this: (slime-setup '(inferior-slime slime-fancy slime-asdf)) works correctly. (I guess I introduced this a while back and just never noticed.) The reason is that inferior-slime and slime-repl (loaded via slime- fancy) are mutually exclusive with respect to their def-slime-selector- method ?r definitions: (def-slime-selector-method ?r -- contrib/inferior-lisp.el "SLIME Read-Eval-Print-Loop." (inferior-slime-switch-to-repl-buffer))) (def-slime-selector-method ?r -- contrib/slime-repl.el "SLIME Read-Eval-Print-Loop." (slime-output-buffer)) However neither slime-fancy (nor just slime-repl) will load without inferior-slime. Either loses with: Symbol's value as variable is void: inferior-slime-mode-map ...and so the natural tendency is to tack on inferior-slime to the end of your slime-setup which results in this bug. Please advise if there's a better way do this. Thanks, Derrell PS, I'm glad you backed out the reversed restart numbering change. Please don't do that.

2 2