ABCL doesn't yet know about the long form of defsetf, used in xref.lisp. Do clispers (the only users of xref in slime) have anything against the patch below? It seems to work in clisp, but I'm no defsetf expert, so I'd rather ask before committing.
Andras
=================================================================== RCS file: /project/slime/cvsroot/slime/xref.lisp,v retrieving revision 1.2 diff -u -r1.2 xref.lisp --- xref.lisp 17 May 2004 00:25:24 -0000 1.2 +++ xref.lisp 17 Sep 2004 20:40:15 -0000 @@ -736,14 +736,14 @@ (:callers (gethash name *callers-database*)) (:readers (gethash name *readers-database*)) (:setters (gethash name *setters-database*)))) -(defsetf callers-list (name &optional (database :callers)) (caller) - `(setf (gethash ,name (case ,database - (:file *file-callers-database*) - (:callees *callees-database*) - (:callers *callers-database*) - (:readers *readers-database*) - (:setters *setters-database*))) - ,caller)) + +(defun (setf callers-list) (val name &optional (database :callers)) + (setf (gethash name (case database + (:file *file-callers-database*) + (:callees *callees-database*) + (:callers *callers-database*) + (:readers *readers-database*) + (:setters *setters-database*))) val))
(defun list-callers (symbol) "Lists all functions which call SYMBOL as a function (function invocation)." @@ -788,9 +788,10 @@ (defun source-file (symbol) "Lists the names of files in which SYMBOL is defined/used." (gethash symbol *source-file*)) -(defsetf source-file (name) (value) - `(setf (gethash ,name *source-file*) ,value))
+(defun (setf source-file) (value name) + (setf (gethash name *source-file*) value)) + (defun clear-tables () (clrhash *file-callers-database*) (clrhash *callers-database*) @@ -807,9 +808,10 @@ (defvar *pattern-caller-type* (make-hash-table :test #'equal)) (defun pattern-caller-type (name) (gethash name *pattern-caller-type*)) -(defsetf pattern-caller-type (name) (value) - `(setf (gethash ,name *pattern-caller-type*) ,value))
+(defun (setf pattern-caller-type) (value name) + (setf (gethash name *pattern-caller-type*) value)) + ;;; Pattern Substitutions (defvar *pattern-substitution-table* (make-hash-table :test #'equal) "Stores general patterns for function destructuring.") @@ -1267,6 +1269,7 @@ database. (For large systems it can take a long time to crunch through the code, so this can save some time.)" (with-open-file (stream filename :direction :output) + (format stream "~&(in-package :pxref)") (format stream "~&(clear-tables)") (maphash #'(lambda (x y) (format stream "~&(setf (source-file '~S) '~S)"
Andras Simon andras@renyi.hu writes:
ABCL doesn't yet know about the long form of defsetf, used in xref.lisp. Do clispers (the only users of xref in slime) have anything against the patch below? It seems to work in clisp, but I'm no defsetf expert, so I'd rather ask before committing.
Is it hard to implement defsetf? I've heard a good principle for Lisp implementors is to take these incompatibilities as a motivation to extend their Lisp rather than contorting programs to use CL subsets. :-)
-Luke
On Mon, 20 Sep 2004, Luke Gorrie wrote:
Andras Simon andras@renyi.hu writes:
ABCL doesn't yet know about the long form of defsetf, used in xref.lisp. Do clispers (the only users of xref in slime) have anything against the patch below? It seems to work in clisp, but I'm no defsetf expert, so I'd rather ask before committing.
Is it hard to implement defsetf? I've heard a good principle for Lisp
Dunno.
implementors is to take these incompatibilities as a motivation to extend their Lisp rather than contorting programs to use CL subsets. :-)
That's a good point.
Anyway, as a temporary measure, I did commit a new version of swank-abcl.lisp which has xref support, but it's commented out. If anyone cares, they could locally patch xref.lisp as in http://article.gmane.org/gmane.lisp.slime.devel/2425, and uncomment the xref section.
BTW, this is as good an occasion as any to ask if there's anybody using slime with abcl. Or if there's anybody who has tried to but met unsurmountable difficulties and gave up on it. I'd love to learn about these! There are some problems that I'm aware of (they mostly have to do with abcl's incomplete Gray streams implementation), but it should be good enough for most kinds of work.
Andras
Andras wrote:
BTW, this is as good an occasion as any to ask if there's anybody using slime with abcl. Or if there's anybody who has tried to but met unsurmountable difficulties and gave up on it. I'd love to learn about these! There are some problems that I'm aware of (they mostly have to do with abcl's incomplete Gray streams implementation), but it should be good enough for most kinds of work.
I'll give it a try later on today and I'll let you know about my experience.
I will be using LATEST from CVS on both ABCL and Slime, is that the right version to use on both?
mb
* Slime is a great, btw. (I commented my ILISP and ELI .emacs code about ~2 months ago, and last week I have simply extracted it into a different file. But given that Slime is moving, in the right directions fast, I do not foresee that I will bring back ILISP and ELI anytime soon ;-)
-
Andras Simon -
Luke Gorrie -
Mike Beedle