Re: [21.5] call-process-region problem

[Katsumi Yamaoka <yamaoka@xxxxxxx>]

>>>>> In [emacs-w3m : No.06092]
>>>>>	Hrvoje Niksic <hniksic@xemacs.org> wrote:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> Although it may be a known problem, I think it is evil that the
>> call-process-region function requires the point to have to be in
>> the beginning or the end position when the 4th arg DELETEP is t.

> It sounds like a bug.

>> It simply erases the buffer's contents when that requirement is
>> not satisfied.

> Why does that happen?  The code looks like it's trying to correctly
> track deletions and insertions by using markers.

It is because call-process-region of that version uses
call-process which inserts the output of a process in the point.
And it deletes the region which is pointed to by markers, after
a process is finished.  Markers function rightly.  Therefore,
they will be erased if the output of a process is inserted
within limits which markers point to.

> It could be that your fix hides another bug, elsewhere in the code,
> but I could be mistaken.  What do you think?

Probably, the problem is simple and lucid.
-- 
Katsumi Yamaoka <yamaoka@jpl.org>