select-active-regions, Emacs not X

select-active-regions, Emacs not X

[Andreas Röhler]

Hi all,

stumble over a real nasty bug, yanking everything back which got into 
the region/highlighted by chance.

See bug#6941: 24.0.50; yank doesn't return latest kill

Reading the news, it seems that it's conceived that way:

"The default handling of clipboard and primary selections has been
changed to conform with other X applications.

*** `select-active-regions' now defaults to t, so active regions set
the primary selection."


IMHO copying everything once got highlighted,  it just the delire :-)

Can't move the cursor without clobbering previous saves.

Emacs is unusable that way.

Well, it's gone after setting `select-active-regions' other than t.

But, please, don't let users run into that mess, don't activate it by 
default.

Well, that feature might be interesting in specific context.
A default value "only shift-selection or mouse-drag" might make sense.

Thanks!

Andreas

--
https://code.launchpad.net/~a-roehler/python-mode
https://code.launchpad.net/s-x-emacs-werkstatt/

Re: select-active-regions, Emacs not X

[Chong Yidong]

Andreas Röhler <andreas.roehler@easy-emacs.de> writes:

> IMHO copying everything once got highlighted,  it just the delire :-)
>
> Can't move the cursor without clobbering previous saves.
>
> Emacs is unusable that way.

During cursor motion, only the primary selection is overwritten, and
only if the region is active and transient mark mode is on.  This does
not affect yanking, since that now uses only the clipboard and the
kill-ring.

The convention on X is that the primary selection is transitory; any
mouse drag or shift selection in any X program changes the primary
selection.  For Emacs to overwrite it when the region is active is not
merely reasonable, it is exactly what other applications do.

All in all, your description of the problem is far too vague.  It is
possible that you have some peculiar customizations that lead to the
unusable behavior you complain about, but it is impossible to know
without a more accurate description.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 30.08.2010 16:18, schrieb Chong Yidong:
> Andreas Röhler<andreas.roehler@easy-emacs.de>  writes:
>
>> IMHO copying everything once got highlighted,  it just the delire :-)
>>
>> Can't move the cursor without clobbering previous saves.
>>
>> Emacs is unusable that way.
>
> During cursor motion, only the primary selection is overwritten, and
> only if the region is active and transient mark mode is on.  This does
> not affect yanking, since that now uses only the clipboard and the
> kill-ring.
>

Doesn't meet my observations.

In the original case cursor motion followed an exchange-point-and-mark 
over a string found.
Afterwards, moving the cursor, any active region went into the 
clipboard. Next yank got it from there.
Whilst the string found and copied has been expected, but was not returned.

Ie got some arbitrary content resp. to the move away from the match-point.

> The convention on X is that the primary selection is transitory; any
> mouse drag or shift selection in any X program changes the primary
> selection.  For Emacs to overwrite it when the region is active is not
> merely reasonable, it is exactly what other applications do.

The big difference IMHO is the cursor-motion-region-setting in emacs.
In X, regions are expressingly marked, while in Emacs it just the 
difference beween point and any previously set mark.

Therefor it not that clear, if an existing region has an editory meaning 
as in X, it might just happen.

>
> All in all, your description of the problem is far too vague.  It is
> possible that you have some peculiar customizations that lead to the
> unusable behavior you complain about, but it is impossible to know
> without a more accurate description.
>

For me the matter is done setting `select-active-regions' not t.
I'll expect others with difficulties here.
Maybe lets postpone the issue.
Should no one else reports trouble, fine.

Thanks all BTW. Great joy always and again with Emacs.

Andreas

Re: select-active-regions, Emacs not X

[Chong Yidong]

Andreas Röhler <andreas.roehler@easy-emacs.de> writes:

> Doesn't meet my observations.
>
> In the original case cursor motion followed an exchange-point-and-mark
> over a string found.  Afterwards, moving the cursor, any active region
> went into the clipboard. Next yank got it from there.  Whilst the
> string found and copied has been expected, but was not returned.
>
> Ie got some arbitrary content resp. to the move away from the match-point.

This description is too vague.  Could you please provide a precise,
step-by-step recipe, starting from `emacs -Q'?

> The big difference IMHO is the cursor-motion-region-setting in emacs.
> In X, regions are expressingly marked, while in Emacs it just the
> difference beween point and any previously set mark.

select-active-regions only takes effect when Transient Mark mode is
enabled, which means the region is explicitly marked.  Again, if this is
not what you see, please provide a step-by-step recipe, starting from
`emacs -Q', to demonstrate the problem.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 30.08.2010 19:35, schrieb Chong Yidong:
> Andreas Röhler<andreas.roehler@easy-emacs.de>  writes:
>
>    
>> Doesn't meet my observations.
>>
>> In the original case cursor motion followed an exchange-point-and-mark
>> over a string found.  Afterwards, moving the cursor, any active region
>> went into the clipboard. Next yank got it from there.  Whilst the
>> string found and copied has been expected, but was not returned.
>>
>> Ie got some arbitrary content resp. to the move away from the match-point.
>>      
> This description is too vague.  Could you please provide a precise,
> step-by-step recipe, starting from `emacs -Q'?
>
>    
>> The big difference IMHO is the cursor-motion-region-setting in emacs.
>> In X, regions are expressingly marked, while in Emacs it just the
>> difference beween point and any previously set mark.
>>      
> select-active-regions only takes effect when Transient Mark mode is
> enabled, which means the region is explicitly marked.  Again, if this is
> not what you see, please provide a step-by-step recipe, starting from
> `emacs -Q', to demonstrate the problem.
>
>
>    

emacs -Q

with empty buffer, insert

foo
(kill-new (buffer-substring-no-properties 1 4)) asdf

eval before "asdf"
push-mark there --thats done in my programm exchange-point-and-mark--
move cursor down - region gets highlighted

yank

==> asdf

IMHO pure cursor move should not send the region to clipboard per default.
OTOH that seems a useful feature for mouse-regions

Thanks

Andreas

Re: select-active-regions, Emacs not X

[Jan Djärv]

I still get foo.

	Jan D.


Andreas Röhler skrev 2010-08-30 20.44:
> Am 30.08.2010 19:35, schrieb Chong Yidong:
>> Andreas Röhler<andreas.roehler@easy-emacs.de> writes:
>>
>>> Doesn't meet my observations.
>>>
>>> In the original case cursor motion followed an exchange-point-and-mark
>>> over a string found. Afterwards, moving the cursor, any active region
>>> went into the clipboard. Next yank got it from there. Whilst the
>>> string found and copied has been expected, but was not returned.
>>>
>>> Ie got some arbitrary content resp. to the move away from the match-point.
>> This description is too vague. Could you please provide a precise,
>> step-by-step recipe, starting from `emacs -Q'?
>>
>>> The big difference IMHO is the cursor-motion-region-setting in emacs.
>>> In X, regions are expressingly marked, while in Emacs it just the
>>> difference beween point and any previously set mark.
>> select-active-regions only takes effect when Transient Mark mode is
>> enabled, which means the region is explicitly marked. Again, if this is
>> not what you see, please provide a step-by-step recipe, starting from
>> `emacs -Q', to demonstrate the problem.
>>
>>
>
> emacs -Q
>
> with empty buffer, insert
>
> foo
> (kill-new (buffer-substring-no-properties 1 4)) asdf
>
> eval before "asdf"
> push-mark there --thats done in my programm exchange-point-and-mark--
> move cursor down - region gets highlighted
>
> yank
>
> ==> asdf
>
> IMHO pure cursor move should not send the region to clipboard per default.
> OTOH that seems a useful feature for mouse-regions
>
> Thanks
>
> Andreas
>
>

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Is transient-mark-mode on? Is the region highlighted after moving down?

Here the region, if transient-mark-mode is on, immediatly goes into the 
clipboard.
And is yanked back by default.

Maybe let's wait if someone else complains this.



Am 30.08.2010 21:27, schrieb Jan Djärv:
> I still get foo.
>
>     Jan D.
>
>
> Andreas Röhler skrev 2010-08-30 20.44:
>> Am 30.08.2010 19:35, schrieb Chong Yidong:
>>> Andreas Röhler<andreas.roehler@easy-emacs.de> writes:
>>>
>>>> Doesn't meet my observations.
>>>>
>>>> In the original case cursor motion followed an exchange-point-and-mark
>>>> over a string found. Afterwards, moving the cursor, any active region
>>>> went into the clipboard. Next yank got it from there. Whilst the
>>>> string found and copied has been expected, but was not returned.
>>>>
>>>> Ie got some arbitrary content resp. to the move away from the 
>>>> match-point.
>>> This description is too vague. Could you please provide a precise,
>>> step-by-step recipe, starting from `emacs -Q'?
>>>
>>>> The big difference IMHO is the cursor-motion-region-setting in emacs.
>>>> In X, regions are expressingly marked, while in Emacs it just the
>>>> difference beween point and any previously set mark.
>>> select-active-regions only takes effect when Transient Mark mode is
>>> enabled, which means the region is explicitly marked. Again, if this is
>>> not what you see, please provide a step-by-step recipe, starting from
>>> `emacs -Q', to demonstrate the problem.
>>>
>>>
>>
>> emacs -Q
>>
>> with empty buffer, insert
>>
>> foo
>> (kill-new (buffer-substring-no-properties 1 4)) asdf
>>
>> eval before "asdf"
>> push-mark there --thats done in my programm exchange-point-and-mark--
>> move cursor down - region gets highlighted
>>
>> yank
>>
>> ==> asdf
>>
>> IMHO pure cursor move should not send the region to clipboard per 
>> default.
>> OTOH that seems a useful feature for mouse-regions
>>
>> Thanks
>>
>> Andreas
>>
>>
>

Re: select-active-regions, Emacs not X

[Miles Bader]

Andreas Röhler <andreas.roehler@online.de> writes:
> IMHO pure cursor move should not send the region to clipboard per default.
> OTOH that seems a useful feature for mouse-regions

If you don't have the tmm region activated, cursor movement shouldn't
affect the selection.  If you _do_ have the tmm region activated, on the
other hand, updating the region via cursor movement _should_ affect the
selection, because that's what the X selection is:  roughly speaking
it's the "latest highlighted thing".

There should be no difference between "tmm-keyboard-regions" and
"mouse-regions" -- they are both methods of selecting text, and both
look the same and behave very similarly in most other ways; to make
mouse regions "special" would simply confuse the user and disadvantage
keyboard users.

Note that other apps (e.g., I tested firefox and chromium) also work
this way: updating a region of selected text with the keyboard sets the
X selection.

-Miles

-- 
The car has become... an article of dress without which we feel uncertain,
unclad, and incomplete.  [Marshall McLuhan, Understanding Media, 1964]

Re: select-active-regions, Emacs not X

[Jan Djärv]

Andreas Röhler skrev 2010-08-31 07.54:
> Is transient-mark-mode on? Is the region highlighted after moving down?

Yes and yes.
>
> Here the region, if transient-mark-mode is on, immediatly goes into the
> clipboard.
> And is yanked back by default.
>

Sounds like your installation is broken.  Check out 
x-cut-buffer-or-selection-value in lisp/term/x-win.el.  Step through it with 
edebug and see what it does.

	Jan D.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 31.08.2010 10:08, schrieb Miles Bader:
> Andreas Röhler<andreas.roehler@online.de>  writes:
>    
>> IMHO pure cursor move should not send the region to clipboard per default.
>> OTOH that seems a useful feature for mouse-regions
>>      
> If you don't have the tmm region activated, cursor movement shouldn't
> affect the selection.  If you _do_ have the tmm region activated, on the
> other hand, updating the region via cursor movement _should_ affect the
> selection, because that's what the X selection is:  roughly speaking
> it's the "latest highlighted thing".
>
> There should be no difference between "tmm-keyboard-regions" and
> "mouse-regions" -- they are both methods of selecting text, and both
> look the same and behave very similarly in most other ways; to make
> mouse regions "special" would simply confuse the user and disadvantage
> keyboard users.
>
> Note that other apps (e.g., I tested firefox and chromium) also work
> this way: updating a region of selected text with the keyboard sets the
> X selection.
>
> -Miles
>
>    

Hi,
delivered a bug report with the issue

bug#6941: 24.0.50; yank doesn't return latest kill

Don't see tmm  activated.

BTW all is fine after setting `select-active-regions' away from t.

So I'm afraid, it just does what it should, if `select-active-regions' is t.

Than question boils down not to set it `t' by default.

Thanks

Andreas

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 31.08.2010 11:38, schrieb Jan Djärv:
>
>
> Andreas Röhler skrev 2010-08-31 07.54:
>> Is transient-mark-mode on? Is the region highlighted after moving down?
>
> Yes and yes.
>>
>> Here the region, if transient-mark-mode is on, immediatly goes into the
>> clipboard.
>> And is yanked back by default.
>>
>
> Sounds like your installation is broken.  Check out 
> x-cut-buffer-or-selection-value in lisp/term/x-win.el.  Step through 
> it with edebug and see what it does.
>
>     Jan D.
>

Did just that last day. See answer to Miles also.

BTW could not figure out the place, where clipboard receives the region. 
Just for curiousity so far.
For me it's done setting `select-active-regions' away from t.

Andreas

Re: select-active-regions, Emacs not X

[David De La Harpe Golden]

On 30/08/10 19:44, Andreas Röhler wrote:

 > emacs -Q
>
> with empty buffer, insert
>
> foo
> (kill-new (buffer-substring-no-properties 1 4)) asdf
>
> eval before "asdf"
> push-mark there --thats done in my programm exchange-point-and-mark--
> move cursor down - region gets highlighted
>
> yank
>

By which you mean C-y?  Or mouse-2 ?  They're not equivalent with the 
recent defaults (quite deliberately).

> ==> asdf

I get foo, at least with yank.

Are you running KDE by any chance?  Its clipboard daemon, 'klipper' has 
a (non-default) setting that actually synchronises the selection 
(primary) and clipboard which might cause the symptoms you describe.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 31.08.2010 23:45, schrieb David De La Harpe Golden:
> On 30/08/10 19:44, Andreas Röhler wrote:
>
> > emacs -Q
>>
>> with empty buffer, insert
>>
>> foo
>> (kill-new (buffer-substring-no-properties 1 4)) asdf
>>
>> eval before "asdf"
>> push-mark there --thats done in my programm exchange-point-and-mark--
>> move cursor down - region gets highlighted
>>
>> yank
>>
>
> By which you mean C-y?  Or mouse-2 ?  They're not equivalent with the 
> recent defaults (quite deliberately).
>
>> ==> asdf
>
> I get foo, at least with yank.
>
> Are you running KDE by any chance?  Its clipboard daemon, 'klipper' 
> has a (non-default) setting that actually synchronises the selection 
> (primary) and clipboard which might cause the symptoms you describe.
>
>
>

Ah, yes, that may be it. Thanks.
So, may Emacs 24 provide for this?

Don't have that problem with Emacs 23...

Re: select-active-regions, Emacs not X

[Chong Yidong]

Andreas Röhler <andreas.roehler@online.de> writes:

>> Are you running KDE by any chance?  Its clipboard daemon, 'klipper'
>> has a (non-default) setting that actually synchronises the selection
>> (primary) and clipboard which might cause the symptoms you describe.
>
> Ah, yes, that may be it. Thanks.
> So, may Emacs 24 provide for this?

I don't think there's anything you can do.  That setting should affect
not only Emacs, but any other X application, causing shift-selection to
copy into the keyboard.  Any user that enables this behavior presumably
wants it.

Re: select-active-regions, Emacs not X

[Stefan Monnier]

>> Ah, yes, that may be it. Thanks.
>> So, may Emacs 24 provide for this?

> I don't think there's anything you can do.  That setting should affect
> not only Emacs, but any other X application, causing shift-selection to
> copy into the keyboard.  Any user that enables this behavior presumably
                ^^^^^^^^
                clipboard
> wants it.


        Stefan

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 02.09.2010 17:15, schrieb Chong Yidong:
> Andreas Röhler<andreas.roehler@online.de>  writes:
>
>    
>>> Are you running KDE by any chance?  Its clipboard daemon, 'klipper'
>>> has a (non-default) setting that actually synchronises the selection
>>> (primary) and clipboard which might cause the symptoms you describe.
>>>        
>> Ah, yes, that may be it. Thanks.
>> So, may Emacs 24 provide for this?
>>      
> I don't think there's anything you can do.  That setting should affect
> not only Emacs, but any other X application, causing shift-selection to
> copy into the keyboard.  Any user that enables this behavior presumably
> wants it.
>
>
>    

Yes, who enables it. Presently it's enabled by default, it's set to 
`always'.
With Emacs 23 the default is nil, that's better.

Also value of `only shift-selection or mouse-drag'
seems a reasonable default, `always' not.

Maybe it's not KDE-related though.

Re: select-active-regions, Emacs not X

[David De La Harpe Golden]

On 02/09/10 17:45, Andreas Röhler wrote:
> Am 02.09.2010 17:15, schrieb Chong Yidong:
>> I don't think there's anything you can do. That setting should affect
>> not only Emacs, but any other X application, causing shift-selection to
>> copy into the keyboard. Any user that enables this behavior presumably
>> wants it.
 >
> Yes, who enables it.

In context, Chong Yidong was talking about the potentially relevant KDE 
klipper setting, not any emacs setting (and meant clipboard not keyboard 
as Stefan already pointed out).

> Maybe it's not KDE-related though.

A screenshot of a possible KDE klipper configuration from my system:

http://i.imgur.com/omaek.png

Is yours different? In particular, is "Synchronise contents of the 
clipboard and the selection" turned on?

[Just in case, auf Deutsch: http://i.imgur.com/4kb97.png
"Synchronise contents of the clipboard and the selection" =>
"Inhalt von Zwischenablage und aktueller Auswahl angleichen"]

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 02.09.2010 23:56, schrieb David De La Harpe Golden:
> On 02/09/10 17:45, Andreas Röhler wrote:
>> Am 02.09.2010 17:15, schrieb Chong Yidong:
>>> I don't think there's anything you can do. That setting should affect
>>> not only Emacs, but any other X application, causing shift-selection to
>>> copy into the keyboard. Any user that enables this behavior presumably
>>> wants it.
> >
>> Yes, who enables it.
>
> In context, Chong Yidong was talking about the potentially relevant 
> KDE klipper setting, not any emacs setting (and meant clipboard not 
> keyboard as Stefan already pointed out).
>
>> Maybe it's not KDE-related though.
>
> A screenshot of a possible KDE klipper configuration from my system:
>
> http://i.imgur.com/omaek.png
>
> Is yours different? In particular, is "Synchronise contents of the 
> clipboard and the selection" turned on?
>
> [Just in case, auf Deutsch: http://i.imgur.com/4kb97.png
> "Synchronise contents of the clipboard and the selection" =>
> "Inhalt von Zwischenablage und aktueller Auswahl angleichen"]
>
>
>

Hi, thanks,

however, think that KDE matter points into the wrong direction.

24 works now so far after setting `select-active-regions' other than 
`always'
23 works right out of the box, as these setting is nil.

All I'm saying is: Probably you will get more bug reports, is the 
default-setting of
`select-active-regions' remains `always'.

Just let the user do that, if wanted.

Andreas

Re: select-active-regions, Emacs not X

[Miles Bader]

Andreas Röhler <andreas.roehler@online.de> writes:
> 24 works now so far after setting `select-active-regions' other than
> always' 23 works right out of the box, as these setting is nil.
>
> All I'm saying is: Probably you will get more bug reports, is the
> default-setting of `select-active-regions' remains `always'.

[It isn't `always' by default, it's `t'; but I guess that's what you mean.]

Why do you think that?  If it `select-active-regions' _isn't_ set to t,
then emacs will operate inconsistently, which will confuse some users
and potentially be a source of bug reports.

Since either setting will probably generate some bug reports, but the
current setting is only a problem when the user is running dubious apps
like klipper, it seems safer to keep the current defaults.

-miles

-- 
Acquaintance, n. A person whom we know well enough to borrow from, but not
well enough to lend to.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 03.09.2010 08:47, schrieb Miles Bader:
> Andreas Röhler<andreas.roehler@online.de>  writes:
>    
>> 24 works now so far after setting `select-active-regions' other than
>> always' 23 works right out of the box, as these setting is nil.
>>
>> All I'm saying is: Probably you will get more bug reports, is the
>> default-setting of `select-active-regions' remains `always'.
>>      
> [It isn't `always' by default, it's `t'; but I guess that's what you mean.]
>
> Why do you think that?  If it `select-active-regions' _isn't_ set to t,
> then emacs will operate inconsistently, which will confuse some users
> and potentially be a source of bug reports.
>    

Well, the default delivered presently is "always"

cus-start.el, line 201

   (select-active-regions killing
                     (choice (const :tag "always" t)
                         (const :tag "only shift-selection or 
mouse-drag" only)
                         (const :tag "off" nil))

"only" would work,

so why not change it to?
23 comes with "nil" as default.

See a unnecessary change of behaviour so far.



> Since either setting will probably generate some bug reports, but the
> current setting is only a problem when the user is running dubious apps
> like klipper, it seems safer to keep the current defaults.
>
> -miles
>
>    

Re: select-active-regions, Emacs not X

[Miles Bader]

Andreas Röhler <andreas.roehler@online.de> writes:
> "only" would work,

No, it wouldn't.  The behavior in `only' mode is inconsistent, in a
confusing way (it does have a basis of course, but that doesn't make its
behavior intuitive for naive users).

Some users (e.g., klipper users) may prefer that inconsistent behavior
for various reasons, but that does not make it desirable in the default
case.

-Miles

-- 
Saa, shall we dance?  (from a dance-class advertisement)

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 03.09.2010 11:00, schrieb Miles Bader:
> Andreas Röhler<andreas.roehler@online.de>  writes:
>    
>> "only" would work,
>>      
> No, it wouldn't.  The behavior in `only' mode is inconsistent, in a
> confusing way (it does have a basis of course, but that doesn't make its
> behavior intuitive for naive users).
>
> Some users (e.g., klipper users) may prefer that inconsistent behavior
> for various reasons, but that does not make it desirable in the default
> case.
>
> -Miles
>
>    

OK, but still doesn't explain why the default of ` select-active-regions'
is changed from

23 nil

to

24 t

The latter caused problems, the former not.

BTW just see there is another variables-default changed

x-select-enable-clipboard

23 nil
24 on

that may cause the bug with KDE 3.5 likewise

maybe is the cause in first order?

Re: select-active-regions, Emacs not X

[Miles Bader]

Andreas Röhler <andreas.roehler@online.de> writes:
> OK, but still doesn't explain why the default of ` select-active-regions'
> is changed from
>
> 23 nil
> to
> 24 t

Because the previous default of `nil' made Emacs inconsistent with other
apps.

There is no perfect default setting that works perfectly for everybody,
but the current default seems pretty good.

-Miles

-- 
Non-combatant, n. A dead Quaker.

Re: select-active-regions, Emacs not X

[Andreas Röhler]

Am 03.09.2010 11:55, schrieb Miles Bader:
> Andreas Röhler<andreas.roehler@online.de>  writes:
>    
>> OK, but still doesn't explain why the default of ` select-active-regions'
>> is changed from
>>
>> 23 nil
>> to
>> 24 t
>>      
> Because the previous default of `nil' made Emacs inconsistent with other
> apps.
>
> There is no perfect default setting that works perfectly for everybody,
> but the current default seems pretty good.
>
> -Miles
>
>    

Hm,  given the fact, primary-selection is a more consistent interface to 
expect,
don't understand the defaults

(defcustom x-select-enable-clipboard t
(defcustom x-select-enable-primary nil

Would revers this settings. But lets wait maybe and see if the issue 
comes up again.