M-= in Dired

M-= in Dired

[Chong Yidong]

In Dired, M-= is bound to dired-backup-diff.  This is no good; there's
no good reason for Dired to override the global binding for M-=, which
is count-words-region.

If there are no objections, I will simply unbind dired-backup-diff, and
improve the behavior of = (dired-diff) so that it uses any existing
backup file as the minibuffer input default, like M-x diff does.

Re: M-= in Dired

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

> If there are no objections, I will simply unbind dired-backup-diff, and
> improve the behavior of = (dired-diff) so that it uses any existing
> backup file as the minibuffer input default, like M-x diff does.

Great.  The binding of `M-=' in dired has annoyed me on several
occasions.

Any chance of removing the re-binding of `up' and `down' so that point
doesn't warp to the beginning of the file names at the same time?
That's also annoying.

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

RE: M-= in Dired

[Drew Adams]

> In Dired, M-= is bound to dired-backup-diff.  This is no good; there's
> no good reason for Dired to override the global binding for M-=, which
> is count-words-region.

Actually, there is no good reason for the global binding of `M-=' to
`count-words-region' either.

It is wasteful of a perfectly good repeatable key for `M-=' to be bound to
`count-words-region', which is not a usefully repeatable command.  You won't be
holding down `M-=' to repeat `count-words-region' multiple times.

It would be far better to bind `count-words-region' to something like `C-x M-=',
and thus keep `M-=' free.

It makes sense to use `M-=' for a repeatable action or perhaps as a prefix key
for grouping things that are mnemonic with an equals sign.  I use `C-=', for
example, as a prefix key for various comparisons:

C-= b         ediff-buffers
C-= d         diff
C-= e, C-= f  ediff-files

An equals sign can be suggestive of comparing things or of calculating/printing
a result, and no doubt of other things as well.  (Emacs sometimes uses `:' for
the showing a result, as in `M-:'.)

The point is that we do not have a very good reason for sacrificing `M-=' to a
one-off command like `count-words-region'.  Sure, that's a good command to bind
to some key, but let's bind it to `C-x M-=' or another key that is not so useful
as either a repeatable key or a prefix key.

> If there are no objections, I will simply unbind dired-backup-diff,
> and improve the behavior of = (dired-diff) so that it uses any existing
> backup file as the minibuffer input default, like M-x diff does.

No objection about that from me.  But I would prefer that we find a different
binding for `count-words-region', for the reasons given above.

up & down keys in Dired [was: M-= in Dired]

[Drew Adams]

> Any chance of removing the re-binding of `up' and `down' so that point
> doesn't warp to the beginning of the file names at the same time?
> That's also annoying.

(Please start a new thread (subject line) when you change the subject.)

You give no reason whatsoever for your proposal, other than that something
annoys you.

I, for one, am not annoyed by up/down, n/p, and C-n/C-p staying at the beginning
of file names.  Quite the contrary.  What is your problem?

If I ever needed to move point vertically in another column, I would just turn
on `wdired' on for the duration.  (You might want to bind `wdired' to a key, to
help you deal with your annoyance.)  But frankly, though I use Dired a lot, I
never seem to have a need to do that.

Re: M-= in Dired

[chad]

On 07 Sep 2012, at 08:19, Drew Adams <drew.adams@oracle.com> wrote:
> 
> It makes sense to use `M-=' for a repeatable action or perhaps as a prefix key
> for grouping things that are mnemonic with an equals sign.  I use `C-=', for
> example, as a prefix key for various comparisons:

Isn't C-= impossible to type into text terminals?

RE: M-= in Dired

[Drew Adams]

> > It makes sense to use `M-=' for a repeatable action or 
> > perhaps as a prefix key for grouping things that are
> > mnemonic with an equals sign.  I use `C-=', for
> > example, as a prefix key for various comparisons:
> 
> Isn't C-= impossible to type into text terminals?

Irrelevant.  I did not propose anything about `C-='.

I mentioned that I, personally, use `C-=' as a prefix key, taking advantage of
the mnemonic value of `='.

The point was that `M-=' could be used as a prefix key (or bound to a repeatable
command).  And that makes more sense than wasting `M-=' on `count-words-region'.

This thread is about `M-='.
And `M-=' is usable in a text terminal.

Re: M-= in Dired

[Philipp Haselwarter]

On Fri, Sep 07 2012 17:01 (@1347030067), Lars Ingebrigtsen wrote:

---8<---[snipped 9 lines]---8<---
> Any chance of removing the re-binding of `up' and `down' so that point
> doesn't warp to the beginning of the file names at the same time?
> That's also annoying.

+1
I don't like that behaviour either.

-- 
Philipp Haselwarter

up & down keys in Dired [was: M-= in Dired]

[Drew Adams]

> > Any chance of removing the re-binding of `up' and `down' so 
> > that point doesn't warp to the beginning of the file names
> > at the same time?  That's also annoying.
> 
> +1
> I don't like that behaviour either.

Once more - you might be unhappy, but you do not say why.

Re: up & down keys in Dired [was: M-= in Dired]

[Philipp Haselwarter]

On Fri, Sep 07 2012 19:14 (@1347038056), Drew Adams wrote:

>> > Any chance of removing the re-binding of `up' and `down' so 
>> > that point doesn't warp to the beginning of the file names
>> > at the same time?  That's also annoying.
>> 
>> +1
>> I don't like that behaviour either.
>
> Once more - you might be unhappy, but you do not say why.

Sometimes I don't want point to jump back to the file name because I
might be reading some other column, say the file sizes. At least C-n and
C-p could run regular next/previous-line. The way it is now it feels
very intrusive.

NB: Once more?
-- 
Philipp Haselwarter

RE: up & down keys in Dired [was: M-= in Dired]

[Drew Adams]

> >> I don't like that behaviour either.
> >
> > Once more - you might be unhappy, but you do not say why.
> 
> Sometimes I don't want point to jump back to the file name because I
> might be reading some other column, say the file sizes. At 
> least C-n and C-p could run regular next/previous-line. The way
> it is now it feels very intrusive.

Thanks for giving a reason.  And a reasonable one.

I don't disagree with the usefulness sometimes of staying in the same column.

I do think that if you start separating the bindings of pairs n/p, C-n/C-p, and
down/up then some people will want one or the other pair to have one or the
other behavior.

Unless we add a user option for that (unlikely), IMO this essentially comes down
to a suggestion that individuals bind whichever keys they like in
`dired-mode-map', which is trivial to do.  IOW, I don't see a good reason to
change the default binding for any one of these pairs in particular.

Coming back to your reason, and FWIW, there are also other ways to provide for
help seeing which date, permissions, etc. belong to which files:

1. I already mentioned toggling `wdired', which is no big deal if you bind it to
a key.  Then anyone can use your `C-n' approach without needing to change
C-n/C-p bindings.  It depends how often you need to use `C-n' that way, I guess.

2. Another answer would be `hl-line' highlighting.  That too can be quickly
toggled on/off.

3. To me, it makes sense for highlighting of a line in Dired to be across the
entire line - like using a ruler with tabular data.  Printed tables often have
rules or zebra striping for just that reason.

But I personally prefer not to use `hl-line' mode all the time in Dired.  I
don't want to always see the current line highlighted.  But I do want to see it
highlighted in some contexts.

So as an alternative to toggling `hl-line' highlighting on/off, I use full-line
highlighting for Dired mouseover and for marked lines.

The suggestion for vanilla Emacs to do likewise was rejected long ago, but I
wouldn't be without it.  I do the same thing for grep output, buffer menu, etc.
You can see the effect for Dired here:
http://www.emacswiki.org/cgi-bin/wiki/DiredPlus

> NB: Once more?

"Once more" referred to my reply: "I'm saying this once again."  I had just
finished replying the same thing to the OP: he gave no reason for his annoyance.
Sorry for any confusion about that.

Re: up & down keys in Dired [was: M-= in Dired]

[Eli Zaretskii]

> From: Philipp Haselwarter <philipp@haselwarter.org>
> Date: Fri, 07 Sep 2012 21:21:03 +0200
> Cc: emacs-devel@gnu.org
> 
> Sometimes I don't want point to jump back to the file name because I
> might be reading some other column, say the file sizes. At least C-n and
> C-p could run regular next/previous-line.

We could use some optional behavior, like "C-x C-n", for that purpose.

(Not that I understand how the fact that point is somewhere on a line
prevents you from reading something else on that line.)

Re: up & down keys in Dired [was: M-= in Dired]

[Richard Stallman]

The main reason for the feature of jumping to the file name
was to help make the names easy to see.  In 1984 putting the cursor
at the start of the name was the only way to do this.

Today maybe some sort of font-lock highlighting would be more effective.

If this is changed, it would be good to have some convenient command
to move point to the start of the file name on the line.  Maybe M-m
should do that.

It might be even better to put point at the start of the name and the
mark at the end of it.

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: M-= in Dired

[Juri Linkov]

> It would be far better to bind `count-words-region' to something like
> `C-x M-=', and thus keep `M-=' free.

Or make `C-x =' a prefix key for a keymap with more keybindings for
counting commands like count-words, count-lines-page,
count-lines-region, count-matches, what-cursor-position,
what-page, what-line.

As for finding a key to rebind `M-=' in Dired, I wonder why couldn't
`C-x v =' call `diff-backup' (or `dired-backup-diff' in Dired) in case when
it can't find a VC backend?

RE: M-= in Dired

[Drew Adams]

> > It would be far better to bind `count-words-region' to 
> > something like `C-x M-=', and thus keep `M-=' free.
> 
> Or make `C-x =' a prefix key
> for a keymap with more keybindings for
> counting commands like count-words, count-lines-page,
> count-lines-region, count-matches, what-cursor-position,
> what-page, what-line.

Sure, `C-x =' or `C-x M-=' or whatever else doesn't waste a repeatable key.

Given the wealth of info that `C-x =' currently returns, even without `C-u', I
don't tend to think of it as counting things, but you could certainly make such
an argument, especially for purposes of grouping keys.

I tend to think of `C-x =' as describing (a character and a position), more than
as counting.  I put it in a class with `C-h f', `C-h v', `C-h c', `C-h k',...
But I also appreciate the `=' mnemonic for the current position.

So I second your suggestion to lump `what-cursor-position' with the other
commands you mention, on the same prefix key.  In general, I'm in favor of that
kind of thing.

Let's free up a great key like `M-=' for better things to come - and _leave_ it
free until we find something that really merits it.

> As for finding a key to rebind `M-=' in Dired, I wonder why couldn't
> `C-x v =' call `diff-backup' (or `dired-backup-diff' in 
> Dired) in case when it can't find a VC backend?

Maybe; why not?  Is there rarely a need to use `diff-backup' when a VC backend
is available?  Dunno.

On the other hand, would it be confusing to a user to not realize which command
(`vc-diff' or `diff-backup') is being used here?  And `C-x v' is generally for
VC stuff - this would present an exception in some contexts.

Re: up & down keys in Dired [was: M-= in Dired]

[Stephen J. Turnbull]

Eli Zaretskii writes:

 > (Not that I understand how the fact that point is somewhere on a line
 > prevents you from reading something else on that line.)

It doesn't prevent me from reading it, it prevents Emacs from reading
it (eg, with M-@ M-w).

Not a big deal to me personally, at least not in a dired buffer.  But
I do understand at least one potential issue with having the current
behavior on multiple keybindings.

Re: M-= in Dired

[Jambunathan K]

Juri Linkov <juri@jurta.org> writes:

>> It would be far better to bind `count-words-region' to something like
>> `C-x M-=', and thus keep `M-=' free.
>
> Or make `C-x =' a prefix key for a keymap with more keybindings for
> counting commands like count-words, count-lines-page,
> count-lines-region, count-matches, what-cursor-position,
> what-page, what-line.

C-x l		count-lines-page

I see that C-x w is empty.  May be `w' can stand for words.

I see that students, academics, writers interested in word counts.  Line
count or page count is not well defined.  They can be "fixed" (as in
"match fixing").

I would probably use C-x = for balancing windows (note the symmetry of
the character and `=' could stand for balancing your accounts) and use
C-x + for counting words (counting is just 1+).

C-x ? can be used for what, if it is at all decided that what-* commands
are worth a key binding.

I would prefer if I could conjure up an image as mnemonic to recall the
exact key sequence to use.
-- 

Re: M-= in Dired

[Juri Linkov]

> If there are no objections, I will simply unbind dired-backup-diff, and
> improve the behavior of = (dired-diff) so that it uses any existing
> backup file as the minibuffer input default, like M-x diff does.

I think it's possible to do this in dired-diff without throwing away
useful functionality of using as default a marked file or a file in
adjacent windows (dwim-target).  The top default value could be
a backup file but remaining defaults could remain the same as before.

PS: Since `diff-backup' uses `diff-latest-backup-file',
would it be more correct to use it in `diff' and `dired-diff'
instead of `file-newest-backup'?

Re: M-= in Dired

[Juri Linkov]

> I think it's possible to do this in dired-diff without throwing away
> useful functionality of using as default a marked file or a file in
> adjacent windows (dwim-target).  The top default value could be
> a backup file but remaining defaults could remain the same as before.

This patch keeps the new functionality of using the latest existing
backup file, and restores the original functionality of using the default
file with the mark on it and adding more defaults for dwim directories:

=== modified file 'lisp/dired-aux.el'
--- lisp/dired-aux.el	2012-09-13 23:42:39 +0000
+++ lisp/dired-aux.el	2012-09-14 23:45:05 +0000
@@ -62,17 +62,31 @@ (defun dired-diff (file &optional switch
 which is the string of command switches for `diff'."
   (interactive
    (let* ((current (dired-get-filename t))
-	  (oldf (file-newest-backup current))
-	  (dir (if oldf (file-name-directory oldf))))
-     (list (read-file-name
-	    (format "Diff %s with%s: "
-		    (file-name-nondirectory current)
-		    (if oldf
-			(concat " (default "
-				(file-name-nondirectory oldf)
-				")")
-		      ""))
-	    dir oldf t)
+	  ;; Get the latest existing backup file.
+	  (oldf (diff-latest-backup-file current))
+	  ;; Get the file at the mark.
+	  (file-at-mark (if (and transient-mark-mode mark-active)
+			    (save-excursion (goto-char (mark t))
+					    (dired-get-filename t t))))
+	  (default-file (or file-at-mark
+			    (and oldf (file-name-nondirectory oldf))))
+	  ;; Use it as default if it's not the same as the current file.
+	  (default (if (not (equal default-file current))
+		       default-file))
+	  (defaults
+	    (if (not (equal (dired-dwim-target-directory)
+			    (dired-current-directory)))
+		(dired-dwim-target-defaults (list current)
+					    (dired-dwim-target-directory)))))
+     (list
+      (read-file-name
+       (format "Diff %s with%s: " current
+	       (if default (format " (default %s)" default) ""))
+       (if default
+	   (dired-current-directory)
+	 (dired-dwim-target-directory))
+       (delq nil (append (list default) defaults))
+       t)
 	   (if current-prefix-arg
 	       (read-string "Options for diff: "
 			    (if (stringp diff-switches)

Re: M-= in Dired

[Chong Yidong]

Juri Linkov <juri@jurta.org> writes:

> This patch keeps the new functionality of using the latest existing
> backup file, and restores the original functionality of using the default
> file with the mark on it and adding more defaults for dwim directories:

Looks fine, thanks.  Don't forget to update the docstring.

Re: M-= in Dired

[Juri Linkov]

> Looks fine, thanks.  Don't forget to update the docstring.

There is a strange situation with `diff-latest-backup-file'.
Its comment says "actually belongs into files.el", but
wouldn't it be appropriate for a function with the `diff-'
prefix to be in files.el?

Re: M-= in Dired

[Chong Yidong]

Juri Linkov <juri@jurta.org> writes:

> There is a strange situation with `diff-latest-backup-file'.  Its
> comment says "actually belongs into files.el", but wouldn't it be
> appropriate for a function with the `diff-' prefix to be in files.el?

I don't think `diff-latest-backup-file' belongs in files.el, but it
ought to be autoloaded if dired-diff uses it.  I did that in trunk.

Frankly, I'm not sure `diff-latest-backup-file' needs to exist.  Its
only purpose is to run the `diff-latest-backup-file' file handler if
there's one, but nothing in the Emacs code tree actually implements that
handler.  So it seems to be a bit of useless cruft.

Re: M-= in Dired

[Stefan Monnier]

> handler.  So it seems to be a bit of useless cruft.

Which seems to argue against auto-loading it ;-)


        Stefan