fill-paragrapch set the buffer to changed when there is none

fill-paragrapch set the buffer to changed when there is none

[Lennart Borgman]

If you do fill-paragraph on a paragraph that is already filled in an 
Elisp buffer there is an undo entry added and the buffer is set to 
modified. Is that correct?

Re: fill-paragrapch set the buffer to changed when there is none

[Mathias Dahl]

> If you do fill-paragraph on a paragraph that is already filled in an
> Elisp buffer there is an undo entry added and the buffer is set to
> modified. Is that correct?

I don't know if it is correct or not, but I have been annoyed by this
now and then.

Re: fill-paragrapch set the buffer to changed when there is none

[Richard Stallman]

    If you do fill-paragraph on a paragraph that is already filled in an 
    Elisp buffer there is an undo entry added and the buffer is set to 
    modified. Is that correct?

It would be better to avoid modifying the buffer.  This has not been
implemented because it isn't trivial to do.

Re: fill-paragrapch set the buffer to changed when there is none

[Lennart Borgman]

Richard Stallman wrote:
>     If you do fill-paragraph on a paragraph that is already filled in an 
>     Elisp buffer there is an undo entry added and the buffer is set to 
>     modified. Is that correct?
>
> It would be better to avoid modifying the buffer.  This has not been
> implemented because it isn't trivial to do.
>   
Would it be possible to just set the modified flag correctly? Can the 
undo list be truncated?

Re: fill-paragrapch set the buffer to changed when there is none

[Paul Pogonyshev]

Lennart Borgman wrote:
> Richard Stallman wrote:
> >     If you do fill-paragraph on a paragraph that is already filled in an 
> >     Elisp buffer there is an undo entry added and the buffer is set to 
> >     modified. Is that correct?
> >
> > It would be better to avoid modifying the buffer.  This has not been
> > implemented because it isn't trivial to do.
> >   
> Would it be possible to just set the modified flag correctly? Can the 
> undo list be truncated?

Yes (at least the first part), but it is not easy to detect when a buffer
is not modified.  If you can tell you are not modifying the buffer, you
can as well not call `insert' and friends at all.

Paul

Re: fill-paragrapch set the buffer to changed when there is none

[Lennart Borgman]

Paul Pogonyshev wrote:
> Lennart Borgman wrote:
>   
>> Richard Stallman wrote:
>>     
>>>     If you do fill-paragraph on a paragraph that is already filled in an 
>>>     Elisp buffer there is an undo entry added and the buffer is set to 
>>>     modified. Is that correct?
>>>
>>> It would be better to avoid modifying the buffer.  This has not been
>>> implemented because it isn't trivial to do.
>>>   
>>>       
>> Would it be possible to just set the modified flag correctly? Can the 
>> undo list be truncated?
>>     
>
> Yes (at least the first part), but it is not easy to detect when a buffer
> is not modified.  If you can tell you are not modifying the buffer, you
> can as well not call `insert' and friends at all.
>
> Paul
>   
Is it not enough just to compare the contents of the filled region 
before and after filling?

Re: fill-paragrapch set the buffer to changed when there is none

[Richard Stallman]

    Would it be possible to just set the modified flag correctly? Can the 
    undo list be truncated?

It might work to fix up the data that way, if it can determine that
the text isn't really changed.  Determining that may not be trivial,
though.