GNU bug report logs - #7604
`loop', PATCH: loop, cl-parse-loop-clause, cl-map-intervals -- being the intervals of

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 7604 in the body.
You can then email your comments to 7604 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: MON KEY <monkey <at> sandpframing.com>
To: bug-gnu-emacs <at> gnu.org
Subject: `loop', PATCH: loop, cl-parse-loop-clause, cl-map-intervals -- being
	the intervals of
Date: Thu, 9 Dec 2010 20:30:03 -0500

Current through BZR-102620 there is a bug around `loop',
`cl-parse-loop-clause', and `cl-map-intervals' which put them in
conflict with the manual (which is unclear w/re intended usage):

 ,----
 | `for VAR being the intervals [of BUFFER] ...'
 |  This clause iterates over all intervals of a buffer with constant
 |  text properties.  The variable VAR will be bound to conses of
 |  start and end positions, where one start position is always equal
 |  to the previous end position.  The clause allows `of', `from',
 |  `to', and `property' terms, where the latter term restricts the
 |  search to just the specified property.  The `of' term may specify
 |  either a buffer or a string.
 `----

(loop for <VAR> being the intervals of property <PROP> from <INT> to <INT>
      collecting <VAR>)

The above works (by accident).

What should work (but doesn't) is:

(loop for <VAR> being the intervals in <BUFFER> of property <PROP>
from <INT> to <INT>
      collecting <VAR>)

The `intervals' feature is an extension to the Common Lisp `loop'
macro but is not in keeping with the syntax.

My understanding of the above is following should work (it doesn't):

(loop for <VAR> being the intervals of <BUFFER> of property <PROP>
from <INT> to <INT>
      collecting tp)

Additionally, the source provides that `in' may be used in the same
manner as `of':

(loop for <VAR> being the intervals in <BUFFER> property <PROP> from
<INT> to <INT>
      collecting tp)

So, I would also expect either of these to work as well (they don't):

(loop for <VAR> being the intervals in <BUFFER> of property <PROP>
from <INT> to <INT>
      collecting tp)

(loop for <VAR> being the intervals of <BUFFER> in property <PROP>
from <INT> to <INT>
      collecting tp)

There should be an additional infix filler word preceding `property'
as their is with the hash-table loop clauses, and he right fix IMO
would be to add a `using' syntax instead of and/or in addition to
testing for `property' which is pretty clearly (of itself) NTRT.
Such an addtion would allow for:

(loop for <VAR> being the intervals in <BUFFER> using property <PROP>
from <INT> to <INT>
      collecting tp)

Regardless, the ttached patch is current w/ BZR-102620. It fixes the
problem and remains backwards compatable with the existing behaviour
in Emacs 23.

rgrep says nothing in core uses this loop idiom so it should be a
fairly clean fix.

--- /ediff25130lIS	2010-12-09 20:06:14.727993830 -0500
+++ /lisp/emacs-lisp/cl-macs.el	2010-12-09 19:54:51.992860000 -0500
@@ -916,17 +916,23 @@
 			  (lambda (,var ,(make-symbol "--cl-var--"))
 			    (progn . --cl-map) nil)
 			  ,buf ,from ,to))))
-
+
 	       ((memq word '(interval intervals))
 		(let ((buf nil) (prop nil) (from nil) (to nil)
 		      (var1 (make-symbol "--cl-var1--"))
 		      (var2 (make-symbol "--cl-var2--")))
-		  (while (memq (car loop-args) '(in of property from to))
-		    (cond ((eq (car loop-args) 'from) (setq from (cl-pop2 loop-args)))
-			  ((eq (car loop-args) 'to) (setq to (cl-pop2 loop-args)))
-			  ((eq (car loop-args) 'property)
-			   (setq prop (cl-pop2 loop-args)))
-			  (t (setq buf (cl-pop2 loop-args)))))
+                  (while (memq (car loop-args) '(in of property from to))
+                    (cond ((eq (car loop-args) 'from) (setq from
(cl-pop2 loop-args)))
+                          ((eq (car loop-args) 'to) (setq to (cl-pop2
loop-args)))
+                          ((eq (car loop-args) 'property) (setq prop
(cl-pop2 loop-args)))
+                          ((memq (car loop-args) '(in of))
+                           (or (and (eq (cadr loop-args) 'property)
+                                    (pop loop-args)
+                                    (setq prop (cl-pop2 loop-args)))
+                               (setq buf (cl-pop2 loop-args))))
+                          (t (or (and (or (stringp (car loop-args))
(bufferp (car loop-args)))
+                                      (setq buf (car loop-args)))
+                                 (error "Expected buffer or string")))))
 		  (if (and (consp var) (symbolp (car var)) (symbolp (cdr var)))
 		      (setq var1 (car var) var2 (cdr var))
 		    (push (list var (list 'cons var1 var2)) loop-for-sets))

--
/s_P\

Message #10 received at 7604 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: MON KEY <monkey <at> sandpframing.com>
Cc: 7604 <at> debbugs.gnu.org
Subject: Re: bug#7604: `loop', PATCH: loop, cl-parse-loop-clause,
 cl-map-intervals -- being the intervals of
Date: Fri, 26 Feb 2016 16:57:48 +1030

MON KEY <monkey <at> sandpframing.com> writes:

> Current through BZR-102620 there is a bug around `loop',
> `cl-parse-loop-clause', and `cl-map-intervals' which put them in
> conflict with the manual (which is unclear w/re intended usage):
>
>  ,----
>  | `for VAR being the intervals [of BUFFER] ...'
>  |  This clause iterates over all intervals of a buffer with constant
>  |  text properties.  The variable VAR will be bound to conses of
>  |  start and end positions, where one start position is always equal
>  |  to the previous end position.  The clause allows `of', `from',
>  |  `to', and `property' terms, where the latter term restricts the
>  |  search to just the specified property.  The `of' term may specify
>  |  either a buffer or a string.
>  `----

[...]

> There should be an additional infix filler word preceding `property'
> as their is with the hash-table loop clauses, and he right fix IMO
> would be to add a `using' syntax instead of and/or in addition to
> testing for `property' which is pretty clearly (of itself) NTRT.
> Such an addtion would allow for:
>
> (loop for <VAR> being the intervals in <BUFFER> using property <PROP>
> from <INT> to <INT>
>       collecting tp)
>
> Regardless, the ttached patch is current w/ BZR-102620. It fixes the
> problem and remains backwards compatable with the existing behaviour
> in Emacs 23.
>
> rgrep says nothing in core uses this loop idiom so it should be a
> fairly clean fix.

Since nothing uses it, and it's a non-standard Emacs extension to loop,
perhaps we should just remove the intervals stuff from cl-loop?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #13 received at 7604 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: MON KEY <monkey <at> sandpframing.com>
Cc: 7604 <at> debbugs.gnu.org
Subject: Re: bug#7604: `loop', PATCH: loop, cl-parse-loop-clause,
 cl-map-intervals -- being the intervals of
Date: Thu, 27 Jun 2019 18:45:22 +0200

MON KEY <monkey <at> sandpframing.com> writes:

> (loop for <VAR> being the intervals in <BUFFER> property <PROP> from
> <INT> to <INT>
>       collecting tp)
>
> So, I would also expect either of these to work as well (they don't):
>
> (loop for <VAR> being the intervals in <BUFFER> of property <PROP>
> from <INT> to <INT>
>       collecting tp)
>
> (loop for <VAR> being the intervals of <BUFFER> in property <PROP>
> from <INT> to <INT>
>       collecting tp)
>
> There should be an additional infix filler word preceding `property'
> as their is with the hash-table loop clauses, and he right fix IMO
> would be to add a `using' syntax instead of and/or in addition to
> testing for `property' which is pretty clearly (of itself) NTRT.

Well, the manual says (if I understand it correctly) that "property"
itself is the keyword here, so in/of isn't necessarily valid, even if it
makes sense.

But nothing uses this stuff (fortunately), so I don't think there's any
point in extending it, so I'm closing this bug report.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.