bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Drew Adams]

This is a regression - loss of a capability that Emacs has had since it
supported a mouse.  How can you classify a regression as "wishlist"?

> > severity 11364 wishlist
> Bug #11364 [emacs] 24.1.50; mouse selection followed by 
> cursor movment to extend selection
> Severity set to 'wishlist' from 'normal'

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Chong Yidong]

"Drew Adams" <drew.adams@oracle.com> writes:

> This is a regression - loss of a capability that Emacs has had since
> it supported a mouse.  How can you classify a regression as
> "wishlist"?

It is not a regression; you are asking for a feature that never existed.

> In Emacs versions prior to Emacs 23 an important feature was the
> ability to extend or reduce the active region by moving the cursor.
> E.g C-SPC followed by C-f.
 
> And it did not matter how the region was activated (how the mark was
> set).  You could, for example, double-click a word or a line or a sexp
> using mouse-1, then use C-f or M-f to extend the region etc.

I checked in Emacs 22: double clicking with the mouse to select a word,
then using C-f, makes the region inactive.  Emacs 24 behaves similarly.

Maybe this is not what you meant.  In that case, I suggest (not for the
first time) giving a step-by-step recipe to reprouduce the exact
behavior you think is problematic.

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Drew Adams]

> > This is a regression - loss of a capability that Emacs has had since
> > it supported a mouse.  How can you classify a regression as
> > "wishlist"?
> 
> It is not a regression; 

I beg to differ; it certainly is.

> you are asking for a feature that never existed.

No, I am asking for the restoration of a feature that _always_ existed, before
Emacs 23.  However, it is unlikely that you will find any mention of this loss
in the NEWS...

As I said in the bug report, I use `delete-selection-mode'.
I should no doubt have emphasized that better - mea culpa.

In Emacs 22 (or prior):

emacs -Q

M-x delete-selection-mode ; turn it on

Double click a word, sexp, line, or what have you.  Or select the region some
other way.  The point is to get an active region.

Use `left', `right', `up', or `down' cursor keys (or `C-f' etc. if you prefer).
The region stays active and is extended (or restricted).

This useful, standard Emacs feature was lost starting with Emacs 23.  Now,
moving the cursor deactivates the region.  It should not, at least not in
`delete-selection-mode'.

> > In Emacs versions prior to Emacs 23 an important feature was the
> > ability to extend or reduce the active region by moving the cursor.
> > E.g C-SPC followed by C-f.
>  
> > And it did not matter how the region was activated (how the mark was
> > set).  You could, for example, double-click a word or a 
> > line or a sexp using mouse-1, then use C-f or M-f to extend the region etc.
> 
> I checked in Emacs 22: double clicking with the mouse to 
> select a word, then using C-f, makes the region inactive.
> Emacs 24 behaves similarly.

Turn on `delete-selection-mode' and you will see that Emacs 24 does not behave
similarly to Emacs 22 (or 21 or...).

I wish Emacs acted like it used to wrt this feature.  Does that make fixing this
fodder for the wishlist?  It should not.

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Drew Adams]

> In Emacs 22 (or prior):
> 
> emacs -Q
> 
> M-x delete-selection-mode ; turn it on
> 
> Double click a word, sexp, line, or what have you.  Or select 
> the region some other way.  The point is to get an active region.
> 
> Use `left', `right', `up', or `down' cursor keys (or `C-f' 
> etc. if you prefer). The region stays active and is extended
> (or restricted).
> 
> This useful, standard Emacs feature was lost starting with 
> Emacs 23.  Now, moving the cursor deactivates the region.
> It should not, at least not in `delete-selection-mode'.

Based on the definitions of "wishlist" and "normal" severity, I reclassified
this easily reproducible regression as "normal".
http://debbugs.gnu.org/Developer.html#severities

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Chong Yidong]

"Drew Adams" <drew.adams@oracle.com> writes:

>> M-x delete-selection-mode ; turn it on
>> 
>> Double click a word, sexp, line, or what have you.  Or select 
>> the region some other way.  The point is to get an active region.
>> 
>> Use `left', `right', `up', or `down' cursor keys (or `C-f' 
>> etc. if you prefer). The region stays active and is extended
>> (or restricted).
>> 
>> This useful, standard Emacs feature was lost starting with 
>> Emacs 23.  Now, moving the cursor deactivates the region.
>> It should not, at least not in `delete-selection-mode'.
>
> Based on the definitions of "wishlist" and "normal" severity, I reclassified
> this easily reproducible regression as "normal".
> http://debbugs.gnu.org/Developer.html#severities

The behavior of Emacs 22 is an unintended and (AFAICT) undocumented
consequence of the delete-selection-mode implementation---basically a
bug, since it had nothing to do with deleting selections.  We might
eventually re-implement this as an optional behavior, but that would be
a new feature (it will certainly not be the default behavior, since it
is very non-standard for graphical applications).

So I do not regard this as a regression, or even a bug.  I'm setting the
severity to wishlist, because there are better things to work on.  Do
not play with the severity---that will not increase the likelihood of
this feature being implemented.

bug#11364: [debbugs-tracker] Processed: severity 11364 wishlist

[Drew Adams]

>> M-x delete-selection-mode ; turn it on
>> 
>> Double click a word, sexp, line, or what have you.  Or select 
>> the region some other way.  The point is to get an active region.
>> 
>> Use `left', `right', `up', or `down' cursor keys (or `C-f' 
>> etc. if you prefer). The region stays active and is extended
>> (or restricted).
>> 
>> This useful, standard Emacs feature was lost starting with 
>> Emacs 23.  Now, moving the cursor deactivates the region.
>> It should not, at least not in `delete-selection-mode'.
> 
> The behavior of Emacs 22 is an unintended and (AFAICT) undocumented
> consequence of the delete-selection-mode implementation---basically a
> bug, since it had nothing to do with deleting selections.  We might
> eventually re-implement this as an optional behavior, but 
> that would be a new feature (it will certainly not be the default
> behavior, since it is very non-standard for graphical applications).
> 
> So I do not regard this as a regression, or even a bug.  I'm 
> setting the severity to wishlist, because there are better things
> to work on.  Do not play with the severity---that will not increase
> the likelihood of this feature being implemented.

You are trying to rewrite history.  Simply stating something does not make it
true.

You claimed that this was NEVER the behavior in Emacs.  Now you admit OK, it
existed in Emacs 22, but that was just a fluke, not intended.  On what basis do
you claim that?

None given, except that there was no specific mention of this in the doc.  The
doc for `delete-selection' mode is minimal.  Have you found anything in the doc
that says that this behavior was NOT intended?

People have been using `delete-selection-mode' for decades.  Do you have ANY
evidence that anyone thought that the way it behaved in this regard was a BUG?
Quite the contrary - users have been taking advantage of this useful feature.
And a bug was filed when you broke it.

In fact, this has been the behavior not just in Emacs 22 but in ALL prior Emacs
versions as well (since `delete-selection-mode' was added).  And there is zero
evidence that this behavior was in any way a mistake, bug, or unintended.  You
give no argument to support your claims.

You play on words, saying that because the traditional behavior "had nothing to
do with deleting selections" it was a bug.  Shame.  Delete-selection mode is not
and has never been only about "deleting selections".

Do you even use `delete-selection-mode'?  I'd guess not, since you had no idea
that you broke this feature.  Did you take a poll of users of
`delete-selection-mode' before breaking it?  Or before retroactively deciding
now that its behavior was a bug that was happily "fixed" by your change that
unknowingly broke it?

Delete-selection mode is not, and was never intended to be, limited to what
might be "standard for graphical applications", any more than the Emacs mouse
was intended to be limited to what is "standard for graphical applications".  

Emacs has always felt free to offer more (or less) than what might currently be
"standard" elsewhere.  The aim has never been to limit Emacs to what is
"standard for graphical applications".  Richard has been very clear about this
from the beginning.  We use standards when, and to the degree that, they are
appropriate for Emacs; they do not use us.

Not to mention that similar behavior is and has always been _standard for Emacs_
in its keyboard handling of the region.  You can still set mark and
extend/restrict the active region by hitting cursor keys, thank goodness.  At
least you broke this Emacs standard behavior only for the case where the region
was made active by the mouse, in delete-selection mode.

Wrt your reclassification: The definition given by GNU for "wishlist" severity
is this:

  "for any feature request, and also for any bugs that are very
   difficult to fix due to major design considerations"

This is not a feature request, no matter what word games you play.  It is a
request to restore the traditional behavior, which you inadvertently broke.

Inadvertently?  Yes, since by your own admission you had no idea that you broke
it.  You even claimed that the traditional behavior never existed.

And you have made no argument that this is "very difficult to fix due to major
design considerations".

You are rewriting history, and apparently trying to redefine "wishlist" as well.
Shame.