insert-file-contents disables undo

insert-file-contents disables undo

[Katsumi Yamaoka]

Hi,

When performing `insert-file-contents' at the end of a buffer,
which is not empty, buffer's undo is disabled.  It happens only
with Emacs 23.

(with-temp-buffer
  (buffer-enable-undo)
  (insert "\n")
  (insert-file-contents "~/.emacs")
  buffer-undo-list)
 => t

I was troubled for months with this problem that sometimes
happened when writing a message draft, and I found the cause of
it today.  It happens when I perform `message-insert-signature'
which uses `insert-file-contents' to insert my ~/.signature file.
It is not only on me, is it?  Could someone repair it?

Thanks in advance.

Re: insert-file-contents disables undo

[Miles Bader]

Katsumi Yamaoka <yamaoka@jpl.org> writes:
> When performing `insert-file-contents' at the end of a buffer,
> which is not empty, buffer's undo is disabled.  It happens only
> with Emacs 23.

Ahhhh, so _that's_ why my mail buffers have undo turned off!

I thought it must be my bug, but apparently not...

Thanks for finding that!  I don't immediately see the reason for the bug
in `insert-file-contents'; that function's handling of the undo list is
a huge festering pile of hair...

-Miles

-- 
x
y
Z!

Re: insert-file-contents disables undo

[Andreas Schwab]

Miles Bader <miles.bader@necel.com> writes:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:
>> When performing `insert-file-contents' at the end of a buffer,
>> which is not empty, buffer's undo is disabled.  It happens only
>> with Emacs 23.
>
> Ahhhh, so _that's_ why my mail buffers have undo turned off!

Should be fixed now.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: insert-file-contents disables undo

[Katsumi Yamaoka]

>>>>> Andreas Schwab wrote:
> Miles Bader <miles.bader@necel.com> writes:

>> Katsumi Yamaoka <yamaoka@jpl.org> writes:
>>> When performing `insert-file-contents' at the end of a buffer,
>>> which is not empty, buffer's undo is disabled.  It happens only
>>> with Emacs 23.
>>
>> Ahhhh, so _that's_ why my mail buffers have undo turned off!

> Should be fixed now.

My long standing melancholy has cleared off at last.  Thanks!

Re: insert-file-contents disables undo

[David Kastrup]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

>>>>>> Andreas Schwab wrote:
>> Miles Bader <miles.bader@necel.com> writes:
>
>>> Katsumi Yamaoka <yamaoka@jpl.org> writes:
>>>> When performing `insert-file-contents' at the end of a buffer,
>>>> which is not empty, buffer's undo is disabled.  It happens only
>>>> with Emacs 23.
>>>
>>> Ahhhh, so _that's_ why my mail buffers have undo turned off!
>
>> Should be fixed now.
>
> My long standing melancholy has cleared off at last.  Thanks!

But that may become undone now.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: insert-file-contents disables undo

[Stefan Monnier]

> Thanks for finding that!  I don't immediately see the reason for the bug
> in `insert-file-contents'; that function's handling of the undo list is
> a huge festering pile of hair...

Hmm... but isn't that pile of hair the immediate reason for the bug?


        Stefan

Re: insert-file-contents disables undo

[martin rudalics]

 >>Thanks for finding that!  I don't immediately see the reason for the bug
 >>in `insert-file-contents'; that function's handling of the undo list is
 >>a huge festering pile of hair...
 >
 > Hmm... but isn't that pile of hair the immediate reason for the bug?

The immediate reason is that I forgot that record_insert would collapse
the insertion done by `insert-file-contents' with other insertions.
AFAICT Andreas fixed that and also the mindless one-armed `if' I wrote.
Thanks, Andreas!

Anyway, the pile of hair is rooted deep inside `insert-file-contents'
which tries to reconcile format decoding, `after-insert-file-functions',
keeping the undo-list short and `point' where it was, and whatever have
you.  `insert-file-contents' presently has about 1200 lines of code,
that's clearly too much for my brain.

martin