Re: bug#5175: 23.1.50; mark deactivated without running deactivate-mark-hook when replying with Gnus

Re: bug#5175: 23.1.50; mark deactivated without running deactivate-mark-hook when replying with Gnus

[Katsumi Yamaoka]

>>>>> In <87aaxqdf9z.fsf@thinkpad.tsdh.de>
>>>>>	Tassilo Horn <tassilo@member.fsf.org> wrote:
> I use this code, to have a 'bar cursor type when the mark is active.

> (defun th-activate-mark-init ()
>   (setq cursor-type 'bar))
> (add-hook 'activate-mark-hook 'th-activate-mark-init)

> (defun th-deactivate-mark-init ()
>   (setq cursor-type 'box))
> (add-hook 'deactivate-mark-hook 'th-deactivate-mark-init)

> This works very good, except when replying to a mail with Gnus using `R'
> in summary.  `R' first yanks the quoted contents of the email I want to
> reply to into the message buffer.  After that, I have a 'bar cursor but
> no active region.

> So it seems that it's somehow possible to deactivate the mark without
> running deactivate-mark-hook, and I think that shouldn't be so.

>>>>> In Bzr-99227  Chong Yidong wrote:
> 2010-01-01  Chong Yidong  <cyd@stupidchicken.com>

> 	* message.el (message-exchange-point-and-mark): Call
> 	exchange-point-and-mark with an argument rather than setting
> 	mark-active by hand (Bug#5175).

This incompatible change makes Gnus not work with old Emacsen, so
I don't merge it to the Gnus trunk, i.e., No Gnus.  Any idea not
to use the argument of `exchange-point-and-mark'?

Re: bug#5175: 23.1.50; mark deactivated without running deactivate-mark-hook when replying with Gnus

[Chong Yidong]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> This incompatible change makes Gnus not work with old Emacsen, so
> I don't merge it to the Gnus trunk, i.e., No Gnus.  Any idea not
> to use the argument of `exchange-point-and-mark'?

An alternative is to avoid using exchange-point-and-mark entirely, like
this:

 (goto-char (prog1 (mark t)
               (set-marker (mark-marker) (point)
                           (current-buffer))))

(Please don't replay to emacsbugs.donarmstrong.com/debbugs.gnu.org AND
emacs-pretest-bug at the same time; sending to the latter creates a new
bug in our database.)

Re: bug#5175: 23.1.50; mark deactivated without running deactivate-mark-hook when replying with Gnus

[Chong Yidong]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> 	* message.el (message-exchange-point-and-mark): Call
>> 	exchange-point-and-mark with an argument rather than setting
>> 	mark-active by hand (Bug#5175).
>
> This incompatible change makes Gnus not work with old Emacsen, so
> I don't merge it to the Gnus trunk, i.e., No Gnus.  Any idea not
> to use the argument of `exchange-point-and-mark'?

I've reworked the change to use the simpler functions from my previous
message; I think this should solve the backward compatibility problem.

Re: bug#5175: 23.1.50; mark deactivated without running deactivate-mark-hook when replying with Gnus

[Katsumi Yamaoka]

>>>>> Chong Yidong wrote:
> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>>> 	* message.el (message-exchange-point-and-mark): Call
>>> 	exchange-point-and-mark with an argument rather than setting
>>> 	mark-active by hand (Bug#5175).
>>
>> This incompatible change makes Gnus not work with old Emacsen, so
>> I don't merge it to the Gnus trunk, i.e., No Gnus.  Any idea not
>> to use the argument of `exchange-point-and-mark'?

> I've reworked the change to use the simpler functions from my previous
> message; I think this should solve the backward compatibility problem.

Thanks for fixing it.  I've verified it also clears the problem:

http://news.gmane.org/group/gmane.emacs.gnus.general/thread=56821