Re: [PATCH v2] files: improve the backup procedure to ensure no data is lost

[Michalis Kokologiannakis]

On Wed 08 Jul 2020 at 20:12, Benno Schulenberg 
<bensberg@telfort.nl> wrote:

> Op 07-07-2020 om 11:54 schreef Michalis Kokologiannakis:
> > Indeed, I had noticed these issues, but I thought to send fixes 
> > for them as
> > separate patches. But I think you are right for the former one; 
> > it would be
> > better to incorporate it into this one.
> >
> > Thoughts on the attached?
>
> Thanks.  Looks good.  Will apply it to master tomorrow.

Thank you.

As for follow-ups, I was considering:

   1) Adding/modifying some comments in the code
   2) A switch to turn the failsafes off
   3) Perhaps taking a failsafe backup only once (when the file 
   is first modified), to avoid the backup's overhead if the file 
   is saved repeatedly.

What do you think?

Thank you,
Michalis

> I will drop the references to the other two bugs (that I added) 
> because
> the patch doesn't prevent that the original file remains 
> truncated when
> the target device is full and under concurrent writing pressure.
>
> (The patch just reduces the severity of such a truncation 
> because nano
> will have saved a copy of the original file to the user's home 
> directory.
> But that will be a small consolation when the file is some 
> configuration
> file that must be there for some system to function.)
>
> Benno