Re: [Nautilus-list] Re: when that call to nautilus_directory_force_reload is not needed

[Yoann Vandoorselaere <yoann mandrakesoft com>]

Alex Larsson <alexl redhat com> writes:

> On Wed, 13 Jun 2001, Darin Adler wrote:
> 
> > On Wednesday, June 13, 2001, at 04:38  PM, Yoann Vandoorselaere wrote:
> >
> > > Ok, I just had a look at the way Nautilus work when, for exemple, two
> > > windows look at the same directory. If you create a file then reload
> > > in one of the window, the other get updated. This is fine with me,
> > > except that we are doing all operations two times (and as told in
> > > a previous mail, this increase linearly each time a new window which
> > > monitor this directory is openened).
> > >
> > > Do you know of any solution that would allow us to do all the operation
> > > one time, but share the result in all the windows ?
> >
> > I'm sorry, I don't understand the question.
> >
> > Today, if you do a refresh in one window, both windows get updated, but
> > the I/O is done only once.
> The problem is not the I/O. 

it's part of the problem.

> The problem is that each view of that
> directory gets a lot of file_changed signals, which leads to every view
> updating it's menus. Due to braindeadness in Bobobo, updating the menus
> takes a lot of time, so performance drops rapidly as you get more views
> displaying the directory.

BTW: I'm working on a problem currently where changed signal are emmited
     before finish_loading get called, the files_changed signal trigger 
     the display_pending_file function which itself trigger a relayout...
     This before *any* file get added. 

     My point of view is that it create latency, the changed signal are
     emmited from nautilus-directory-async.c in function that trigger file
     info / file list / uri activation / mime list / top left / etc...
     
     I wonder why things like uri activation have to emit this signal,
     (removed it, seem nothing broke)... But if we can't do that, we might
     disconnect the signal handler as soon as we finish loading a directory,
     then reconnect to them in finish_loading...

     WDYT ?

-- 
Yoann Vandoorselaere | An engineer from NVidia, while asking him to release
MandrakeSoft         | cards specs said : - "Actually, we do write our drivers
                     | without documentation."