Re: [h-e-w] Re: Offline files and delayed writing

[Jason Rumney]

Brian Elmegaard wrote:

> Stephen Leake <address@hidden> writes:
>
>  
>
> > > directory is updated. This fools emacs to think that the file has been
> > > saved outside by another application. 
> > >      
> > It has!
> >    
>
> Yes and no, imho. That the file system is slow in saving a file does
> not mean that some other application has done the save. Does it?
>
>  
Yes, in this case it does. The function Emacs calls to save the file is
a blocking function. Although disk caching is happening in the
background, so it is not quite as simple as "when the function returns
the file is safely stored on disk",The contract between the operating
system and applications that call that function says that the file
should appear to the application to be stored on disk when the function
returns.

> This could be an option but I like auto-revert for output files. Would
> it be more "correct" to have auto-revert as buffer-local?  
>  
Yes. But Emacs will still detect the file as "changed" when you try to
save it, or perform certain other operations, and prompt you as to
whether you want your edited version or the "new" version on disk.

> How does auto-revert check if it is emacs itself or another that has
> saved a file?
It doesn't. All it knows is that the file has been modified since the
last open or save that Emacs made.

> Does it keep an internal variable for the time of last
> save inside emacs?
Basically, yes.

> Can this store the time after the compression
> instead?
>  
How do you propose to do that?