NetAccess::statInternal and showing job progress

[David Faure]

On Thursday 22 July 2010, Sebastian Kügler wrote:
> On Thu July 22 2010 10:17:41 David Faure wrote:
> > On Tuesday 20 July 2010, Chani wrote:
> > > On July 20, 2010 11:32:35 David Faure wrote:
> > > > On Tuesday 20 July 2010, Aaron J. Seigo wrote:
> > > > > On July 20, 2010, you wrote:
> > > > > > On Friday 16 July 2010, Aaron J. Seigo wrote:
> > > > > > > is there any pressing reason to show the progress info for
> > > > > > > stat/exist checks on remote files? i'd like to commit this and
> > > > > > > backport if there is consensus.
> > > > > > 
> > > > > > I object. This is very useful for slow (or even hanging)
> > > > > > connections.
> > > > > 
> > > > > for file transfers, i agree. but for stats and existence checking,
> > > > > what's the use there? other than for debugging, which i really
> > > > > don't think is a good excuse for showing such things to the user
> > > > > every single time, i can't see the benefit to sending job
> > > > > notifications about checking for file details.
> > > > > 
> > > > > stats and existence checking are almost never something the user is
> > > > > actively triggering (from their POV), and so the desktop starts
> > > > > sounding very "chatty" telling them about all kinds of internal
> > > > > details of things they haven't triggered (again, from their POV)
> > > > > and which they really don't care about.
> > > > 
> > > > Again, if it takes <1s, then yes, I agree that they don't need to
> > > > care. But if it takes 10 minutes because the server isn't
> > > > responding, and meanwhile (without any kind of progress info) they
> > > > don't know if the image got saved or not, then this would be a very
> > > > severe bug, leading to data loss, potentially (you think a file
> > > > exists, and it does not).
> > > 
> > > erm, but this wasn't about save, it was about stat :)
> > 
> > ... which is usually done just before saving, to warn in case you're
> > about to overwrite an existing file.
> 
> Still, "I'm checking if there's a file I'd be overwriting" is not something
> that the user needs to know. We might need feedback somewhere, but that
> should be "saving your file now" (and covering the complete process:
> stat'ing, transferring and saving), anything with finer granularity is not
> useful IMO. Likely, as Save is often fire and forget, we'd only need a
> warning if the file wasn't saved in the end, maybe a confirmation that the
> file is now on disk and you can happily pull the cable, or whatever.

I totally agree. However this means making a composite KJob that takes care of 
both checking and writing, and showing progress info for that kjob.

It does NOT mean hiding the progress info during a NetAccess::stat and leaving 
it at that.

-> yet another reason to use proper jobs rather than netaccess.

-- 
David Faure, faure at kde.org, http://www.davidfaure.fr
Sponsored by Nokia to work on KDE, incl. Konqueror (http://www.konqueror.org).