emacs-unicode-2: global-auto-revert-mode doesn't work properly

emacs-unicode-2: global-auto-revert-mode doesn't work properly

[William Xu]

Hi, folks, 

I've set global-auto-revert-mode to t, and there are two emacs opening
the same file, say `foo'. When one of them *inserts*(deletion doesn't
have this problem, strangely..) something and saves foo, then in the
other emacs's foo buffer, it prompts (which shouldn't happend as
global-auto-revert-mode is turned on),

    foo changed on disk; really edit the buffer? (y, n, r or C-h)

Typing either n or r seems no effects, it'll keep prompting. Typing y
removes the prompt, though.

emacs version: GNU Emacs 23.0.0.1 (powerpc-unknown-linux-gnu) of
2006-06-05 (Latest cvs checkout seems still has this problem)

-- 
William

You can't take it with you -- especially when crossing a state line.

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[Kenichi Handa]

In article <87u06xov1q.fsf@www.williamxu.com>, William Xu <william.xwl@gmail.com> writes:

> I've set global-auto-revert-mode to t, and there are two emacs opening
> the same file, say `foo'. When one of them *inserts*(deletion doesn't
> have this problem, strangely..) something and saves foo, then in the
> other emacs's foo buffer, it prompts (which shouldn't happend as
> global-auto-revert-mode is turned on),

>     foo changed on disk; really edit the buffer? (y, n, r or C-h)

> Typing either n or r seems no effects, it'll keep prompting. Typing y
> removes the prompt, though.

I think emacs-unicode-2 doesn't change anything about that
feature.  Don't you see the same problem in CVS HEAD?

---
Kenichi Handa
handa@m17n.org

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[William Xu]

Kenichi Handa <handa@m17n.org> writes:

> In article <87u06xov1q.fsf@www.williamxu.com>, William Xu <william.xwl@gmail.com> writes:
>
>> I've set global-auto-revert-mode to t, and there are two emacs opening
>> the same file, say `foo'. When one of them *inserts*(deletion doesn't
>> have this problem, strangely..) something and saves foo, then in the
>> other emacs's foo buffer, it prompts (which shouldn't happend as
>> global-auto-revert-mode is turned on),
>
>>     foo changed on disk; really edit the buffer? (y, n, r or C-h)
>
>> Typing either n or r seems no effects, it'll keep prompting. Typing y
>> removes the prompt, though.
>
> I think emacs-unicode-2 doesn't change anything about that
> feature.  Don't you see the same problem in CVS HEAD?

Nope... actually i used to be using emacs-snapshot, things just work
fine. I turned to emacs-unicode for gbk support recently, and met that
problem.

-- 
William

Wonderful day.  Your hangover just makes it seem terrible.

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[Kenichi Handa]

In article <87hd2ua7gi.fsf@williamxwl.com>, William Xu <william.xwl@gmail.com> writes:

> Kenichi Handa <handa@m17n.org> writes:
>> In article <87u06xov1q.fsf@www.williamxu.com>, William Xu <william.xwl@gmail.com> writes:
>> 
>>> I've set global-auto-revert-mode to t, and there are two emacs opening
>>> the same file, say `foo'. When one of them *inserts*(deletion doesn't
>>> have this problem, strangely..) something and saves foo, then in the
>>> other emacs's foo buffer, it prompts (which shouldn't happend as
>>> global-auto-revert-mode is turned on),
>> 
>>> foo changed on disk; really edit the buffer? (y, n, r or C-h)
>> 
>>> Typing either n or r seems no effects, it'll keep prompting. Typing y
>>> removes the prompt, though.
>> 
>> I think emacs-unicode-2 doesn't change anything about that
>> feature.  Don't you see the same problem in CVS HEAD?

> Nope... actually i used to be using emacs-snapshot, things just work
> fine. I turned to emacs-unicode for gbk support recently, and met that
> problem.

Ah, I found that this bug happens when you insert non-ASCII
characters, and I found the reason why that causes the
problem only in emacs-unicode-2 (difference in REPLACE
handling in Finsert_file_contents; HEAD uses insert_1 but
emacs-unicode-2 uses insert_from_buffer).  I'm now working
on fixing it.  Please wait for a while.

---
Kenichi Handa
handa@m17n.org

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[William Xu]

Kenichi Handa <handa@m17n.org> writes:

> Ah, I found that this bug happens when you insert non-ASCII
> characters,

Smart guess ! But seems it makes no difference here.. 

> and I found the reason why that causes the
> problem only in emacs-unicode-2 (difference in REPLACE
> handling in Finsert_file_contents; HEAD uses insert_1 but
> emacs-unicode-2 uses insert_from_buffer).  I'm now working
> on fixing it.  Please wait for a while.

Still progress though ! :-)

-- 
William

Peterson's Admonition:
	When you think you're going down for the third time --
	just remember that you may have counted wrong.

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[Kenichi Handa]

I've just reached to my todo-list entry about this matter.

In article <87k67jzvh3.fsf@www.williamxu.com>, William Xu <william.xwl@gmail.com> writes:

> Kenichi Handa <handa@m17n.org> writes:
> > Ah, I found that this bug happens when you insert non-ASCII
> > characters,

> Smart guess ! But seems it makes no difference here.. 

> > and I found the reason why that causes the
> > problem only in emacs-unicode-2 (difference in REPLACE
> > handling in Finsert_file_contents; HEAD uses insert_1 but
> > emacs-unicode-2 uses insert_from_buffer).  I'm now working
> > on fixing it.  Please wait for a while.

> Still progress though ! :-)

insert_1 can be called with the arg PREPARE zero, but as
insert_from_buffer and insert_from_buffer_1 don't have such
an arg, prepare_to_modify_buffer is called unconditionally,
which leads to the call of
ask-user-about-supersession-threat, which makes
global-auto-revert-mode don't work.

I've fixed that by binding buffer-file-name to nil
temporarily.

---
Kenichi Handa
handa@m17n.org

Re: emacs-unicode-2: global-auto-revert-mode doesn't work properly

[William Xu]

Kenichi Handa <handa@m17n.org> writes:

[...]

> insert_1 can be called with the arg PREPARE zero, but as
> insert_from_buffer and insert_from_buffer_1 don't have such
> an arg, prepare_to_modify_buffer is called unconditionally,
> which leads to the call of
> ask-user-about-supersession-threat, which makes
> global-auto-revert-mode don't work.
>
> I've fixed that by binding buffer-file-name to nil
> temporarily.

Hmm, that did fix it, in latest checkout. Thanks ! 

-- 
William

Marijuana will be legal some day, because the many law students
who now smoke pot will someday become congressmen and legalize
it in order to protect themselves.
		-- Lenny Bruce