On Sat, Oct 8, 2016 at 3:33 PM, infirit <infirit@xxxxxxxxx> wrote: > Op 10/08/2016 om 11:52 PM schreef Laurence Gonsalves: >> Is it necessary to use timeout_add? I noticed that I can call >> add_item before starting the main loop, and it seems to work once the >> main loop is started. I'm not sure if that's relying on undocumented >> behavior, however. > > There is no undocumented behaviour, many parts of Gtk and GLib require a > mainloop to function properly, GtkRecentManager is one of them and > sending/receiving signals is another. Sorry if I wasn't clear. I wasn't talking about starting the mainloop. I was talking about calling add_item *before* starting the mainloop rather than scheduling the add_item calls from an event handler. Doing this seems to work, but I'm not sure if it's guaranteed to work or if it's relying on undocumented behavior. > You could do that but it relies, imo, on implicit behaviour that if no > higher tasks are running it should quit. My approach would be that you > explicitly call Gtk.main_quit when you are done processing the files. .... > def on_timeout(manager, files): > for path in files: > f = Gio.File.new_for_path(path) > exists = f.query_exists() > if exists: > res = manager.add_item(f.get_uri()) > print(res) > Gtk.main_quit() I tried this and it didn't work. It looks like main_quit terminates the main loop immediately, before the add_item requests even have a chance to do their job. > My timeout was five seconds, but it could be one milliseconds so it has > no noticeable delay when you run it. Is there any advantage to using a timeout over idle_add? If a timeout must be used, why not use a zero timeout? > [1] https://lazka.github.io/pgi-docs/#Gio-2.0/interfaces/File.html Thanks for that. I'll update my code to use it. _______________________________________________ gtk-list mailing list gtk-list@xxxxxxxxx https://mail.gnome.org/mailman/listinfo/gtk-list
