[Nautilus-list] Re: [PATCH]: avoid directory reloading when not needed.

[Yoann Vandoorselaere <yoann mandrakesoft com>]

Darin Adler <darin bentspoon com> writes:

> On Thursday, June 14, 2001, at 02:13  PM, Yoann Vandoorselaere wrote:
> 
> > What the patch does currently is that it make reloading
> > happen either :
> >
> > - when the refresh button is used.
> > - when a new window is opened.
> >
> > For the second case, it would be good to be able to know
> > if a window already display the directory.
> >
> > Is the patch clean / right enough to commit ?

Something that I forgotten to tell is that new file still appear
when retyping the directory filename even thought it's already
loaded... Do you know why ?

> Hmm. Even though I said "you could try making that change", now that I
> see the actual patch it doesn't seem right to me.
> 
> If someone types a new location into an existing window, I don't think
> the behavior should be any different from opening a new window. In
> either case,
>   if the new location happens to be cached, we need to do a new load
> and not rely on old cached data unless we have synchronization like
> FAM.

FAM has drawback. 
One of them is that it is currently not widely used. And another
stuff that I don't like is that it is depending on portmap.

The current behavior seem really broken to me.

> A particularly bad case of this has to do with going up from your home
> directory, I seem to recall. There were cases where new files just
> didn't show up without a refresh. We can dig around the bug database
> to find this specific awful case. I'm pretty sure this change
> reintroduces the bug.

The solution to this is to issue a reload if the directory is
already monitored. Is there a way to do that ?

Another thing is that I never ever seen Nautilus cache > 1 directory 
at a time.

And the third thing is that even thought the directory is cached,
new file appear on reloading it from the cache (at least it should
be from the cache) (and I really can't understand this one).

> Maybe some of the other folks in the Nautilus community can weigh in
> on this. If there's a consensus among enough of you guys that we
> should do this change, I'll approve it. But I personally think it's
> not an improvement. It brings back the old bug we had where you see
> cached data that confused.

I can't reproduce this bug in my current tree.

> I'm sorry that I encouraged you to write this patch. As I said before: "I'
> d prefer to just make the FAM change though, and not waste time
> optimizing things for the non-FAM case." The FAM change will not
> reintroduce the bug,
>   and it's incredibly simple to do.

Ok. 

BTW : can you look at the mail where I talk about files_changed signal  
      emmited to soon before finish_loading is called and causing re-
      layout (which slow the whole process). I need help on this one...

-- 
Yoann Vandoorselaere | "We've heard that a million monkeys at a million
MandrakeSoft         | keyboards could produce  the Complete Works of
                     | Shakespeare; now, thanks to the Internet, we know  this
                     | is not true."               Robert Wilensky, University
                     | of California