A few questions about occur-mode

A few questions about occur-mode

[Juanma Barranquero]

1.- Would anyone object to making some occur functions (namely
`occur-mode-goto-occurrence', `occur-mode-display-occurrence' and
`occur-mode-goto-occurrence-other-window') to accept an optional
argument and do a (recenter arg) after moving the point? My main use of
occur is:

M-x occur "whatever"
C-x o
C-n C-n ... C-o    # see something
C-n C-n ... C-o    # again...

and it's very handy to put that "something" at the beginning of the
window.

2.- Why are the functions called occur-mode-* instead of simply occur-*?

3.- Do `occur-next' and `occur-prev' do something, or are they remnants
of some long-deleted code, a half-implemented idea, or what? AFAICS, the
'occur-point property is only referenced in `occur-next', `occur-prev'
and `sh-mark-line' (in progmodes/sh-script.el).

4.- IMO it'd be nice to have bindings in occur-mode to do C-n C-o and
C-p C-o (I'd put them in M-n, M-p instead of occur-{next|prev}, but
C-M-n and C-M-p would do fine too). Useful, or something to be relegated
to my .emacs? :-)

Thanks,


                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Richard Stallman]

    1.- Would anyone object to making some occur functions (namely
    `occur-mode-goto-occurrence', `occur-mode-display-occurrence' and
    `occur-mode-goto-occurrence-other-window') to accept an optional
    argument and do a (recenter arg) after moving the point? My main use of
    occur is:

These functions are part of a series that exist in many different
parts of Emacs (including Dired, for instance).  So the decision
of whether to change them in this way calls for more thought.

For now, please don't.

    4.- IMO it'd be nice to have bindings in occur-mode to do C-n C-o and
    C-p C-o (I'd put them in M-n, M-p instead of occur-{next|prev}, but
    C-M-n and C-M-p would do fine too). Useful, or something to be relegated
    to my .emacs? :-)

If we do this, we should do it in all of those modes.  This too calls for
more though and study.

    2.- Why are the functions called occur-mode-* instead of simply occur-*?

No particular reason.

    3.- Do `occur-next' and `occur-prev' do something, or are they remnants
    of some long-deleted code, a half-implemented idea, or what? AFAICS, the
    'occur-point property is only referenced in `occur-next', `occur-prev'
    and `sh-mark-line' (in progmodes/sh-script.el).

I don't remember anything about that.

Re: A few questions about occur-mode

[Juanma Barranquero]

On Sat, 21 Sep 2002 15:39:38 -0400
Richard Stallman <rms@gnu.org> wrote:

> For now, please don't.

Ok. Still, it´d be useful to give it some though as you suggest.

> If we do this, we should do it in all of those modes.  This too calls for
> more though and study.

Ok. My .emacs for now, then.

> No particular reason.

Wouldn´t then be logical to rename them? AFAIK, user functions for a
mode FOO are usually FOO-func1, FOO-func2, not FOO-mode-func1,
FOO-mode-func2. I don´t think changing them would cause trouble, as
users are probably using them throug bindings in occur-mode and not
directly with M-x.

> I don't remember anything about that.

Any idea, anyone?

Thanks,

-- 
Juanma Barranquero <lektu@terra.es>

Re: A few questions about occur-mode

[John Paul Wallington]

Juanma Barranquero <lektu@terra.es> wrote:

>> I don't remember anything about that.
>
> Any idea, anyone?

`occur-next' and `occur-prev' work in the EMACS_21_1_RC branch.

In HEAD they could be changed to use `occur-match' property instead of
`occur-point', or `occur-engine' could add `occur-point' properties to
its output.  Which is better ?

-- 
John Paul Wallington

Re: A few questions about occur-mode

[Richard Stallman]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 355 bytes --]

    Wouldn´t then be logical to rename them? AFAIK, user functions for a
    mode FOO are usually FOO-func1, FOO-func2, not FOO-mode-func1,
    FOO-mode-func2. I don´t think changing them would cause trouble, as
    users are probably using them throug bindings in occur-mode and not
    directly with M-x.

Yes, that renaming is ok if you want to do it.

Re: A few questions about occur-mode

[Juanma Barranquero]

On Mon, 23 Sep 2002 00:45:01 +0100, John Paul Wallington <jpw@shootybangbang.com> wrote:

> `occur-next' and `occur-prev' work in the EMACS_21_1_RC branch.

You're right. Somewhere in the big revamp of occur this functionality
was lost somehow.

> In HEAD they could be changed to use `occur-match' property instead of
> `occur-point', or `occur-engine' could add `occur-point' properties to
> its output.  Which is better ?

Using occur-match seems more economical, but then `occur-next' and
`occur-prev' should have to be modified. The occur-point property was
set in just one character at the end of the match; with occur-match, the
entire match is propertized and occur-{next|prev} would first jump to
the start of the match, then to the end. Not that is difficult, of
course. sh-script.el should have to be modified too.

OTOH, adding back occur-point is a one- or two- line change:

Index: replace.el
===================================================================
RCS file: /cvs/emacs/lisp/replace.el,v
retrieving revision 1.159
diff -u -2 -r1.159 replace.el
--- replace.el	18 Sep 2002 16:03:58 -0000	1.159
+++ replace.el	23 Sep 2002 11:03:55 -0000
@@ -822,5 +822,6 @@
 						`(font-lock-face ,match-face)))
 					     curstring)
-			(setq start (match-end 0))))
+                        (setq start (match-end 0))
+                        (put-text-property start (1+ start) 'occur-point t curstring)))
 		    ;; Generate the string to insert for this match
 		    (let* ((out-line


                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Juanma Barranquero]

On Mon, 23 Sep 2002 13:04:54 +0200, Juanma Barranquero <lektu@terra.es> wrote:

> OTOH, adding back occur-point is a one- or two- line change:

Well, in fact is more like it:

 						`(font-lock-face ,match-face)))
 					     curstring)
+                        (put-text-property (match-beginning 0) (1+ (match-beginning 0)) 'occur-point t curstring)
                        (setq start (match-end 0))))
 		    ;; Generate the string to insert for this match
 		    (let* ((out-line


because otherwise there could be problems with matches at the end of a
line. This leaves the point at the beginning of a match after M-n or M-p,
but that seems preferable IMHO.

Opinions?


                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Richard Stallman]

    In HEAD they could be changed to use `occur-match' property instead of
    `occur-point', or `occur-engine' could add `occur-point' properties to
    its output.  Which is better ?

I think it is cleaner to use just one property for both purposes.

Re: A few questions about occur-mode

[John Paul Wallington]

Juanma Barranquero <lektu@terra.es> wrote:

> because otherwise there could be problems with matches at the end of a
> line. This leaves the point at the beginning of a match after M-n or M-p,
> but that seems preferable IMHO.
>
> Opinions?

If we put `occur-point' properties in `occur-engine' output because
of compatibility maybe they should go at the end of each match for
the same reason?   We could acomplish that by sticking a space on the
end of each line's string before propertizing it:

Index: replace.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/replace.el,v
retrieving revision 1.159
diff -u -r1.159 replace.el
*** replace.el	18 Sep 2002 16:03:58 -0000	1.159
--- replace.el	23 Sep 2002 20:46:42 -0000
***************
*** 804,811 ****
  		    (setq lines (+ lines (1- (count-lines origpt endpt))))
  		    (setq marker (make-marker))
  		    (set-marker marker matchbeg)
! 		    (setq curstring (buffer-substring begpt
! 						      (line-end-position)))
  		    ;; Depropertize the string, and maybe
  		    ;; highlight the matches
  		    (let ((len (length curstring))
--- 804,813 ----
  		    (setq lines (+ lines (1- (count-lines origpt endpt))))
  		    (setq marker (make-marker))
  		    (set-marker marker matchbeg)
! 		    (setq curstring (concat
! 				     (buffer-substring begpt
! 						       (line-end-position))
! 				     " "))
  		    ;; Depropertize the string, and maybe
  		    ;; highlight the matches
  		    (let ((len (length curstring))
***************
*** 821,826 ****
--- 823,834 ----
  					      (when match-face
  						`(font-lock-face ,match-face)))
  					     curstring)
+ 			;; `occur-point' property is used by occur-next and
+ 			;; occur-prev to move between matches.
+ 			(put-text-property (match-end 0)
+ 					   (1+ (match-end 0))
+ 					   'occur-point t 
+ 					   curstring)
  			(setq start (match-end 0))))
  		    ;; Generate the string to insert for this match
  		    (let* ((out-line

-- 
John Paul Wallington

Re: A few questions about occur-mode

[Juanma Barranquero]

On Mon, 23 Sep 2002 11:59:53 -0400, Richard Stallman <rms@gnu.org> wrote:

> I think it is cleaner to use just one property for both purposes.

Ok. I've modified `occur-next' and `occur-prev' to use 'occur-match.

Still, someone knowledgeable on progmodes/sh-script.el should redo
`sh-learn-buffer-indent' and friends.


                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Juanma Barranquero]

On Mon, 23 Sep 2002 21:53:52 +0100, John Paul Wallington <jpw@shootybangbang.com> wrote:

> If we put `occur-point' properties in `occur-engine' output because
> of compatibility maybe they should go at the end of each match for
> the same reason?

Richard has suggested using just one property, so I've chosen
'occur-match, which seems the cleaner method.  The fixes for `occur-next'
and `occur-prev' are pretty straightforward, and now they work as
previously did, AFAICS.


                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Juanma Barranquero]

On Tue, 24 Sep 2002 10:43:57 +0200, Juanma Barranquero <lektu@terra.es> wrote:

> Still, someone knowledgeable on progmodes/sh-script.el should redo
> `sh-learn-buffer-indent' and friends.

Ok, I've done a first cut at fixing `sh-learn-buffer-indent' and it
seems to be working. At least it pops a buffer that you can use with M-n,
M-p and C-c C-c to get to the right places.

But as I had no previous experience with sh-script.el, I'd suggest
someone who does to take a look at it.

Thanks,

                                                           /L/e/k/t/u

Re: A few questions about occur-mode

[Richard Stallman]

    Ok. I've modified `occur-next' and `occur-prev' to use 'occur-match.

Thank you.

Re: A few questions about occur-mode

[Richard Stallman]

    > Still, someone knowledgeable on progmodes/sh-script.el should redo
    > `sh-learn-buffer-indent' and friends.

    Ok, I've done a first cut at fixing `sh-learn-buffer-indent' and it
    seems to be working. At least it pops a buffer that you can use with M-n,
    M-p and C-c C-c to get to the right places.

Thank you.

    But as I had no previous experience with sh-script.el, I'd suggest
    someone who does to take a look at it.

I'd be surprised if any of us knows that code better than you do.