On Fri, Dec 16, 2011 at 10:32 PM, John Lindgren <john.lindg...@aol.com> wrote: > Steve, > > I use the "time" shell command and wait for my CPU meter to drop back to > idle before quitting the test program, giving something like this: > > time ./list-test > > real 0m31.719s > user 0m29.168s > sys 0m0.023s
GtkTreeView should definitely be usable with millions of rows in the data model. Remember that "should" is a word with many implications, things that GtkTreeView should be doing include avoiding requesting the size for every row in the model. And of course your GtkTreeModel implementation should not load all actual data but instead implement the interface in a way that data is only loaded on demand (one should be able to tell how large, how many rows without actually pulling the data for every row). A good starting point would be to open gtktreeview.c and start figuring ways to avoid the loop which currently itterates over the entire data store and finds the perfect size of the treeview. A good technique for this would be to synchronously request the size for all visible rows (which is currently done), plus request the size of 1 or 2 visible page sizes of previous and following rows. After that one might check the width and height requests of only one out of every 10 thousand following rows (assuming a 5 million row treeview or such, make a reasonable calculation), and estimate the total size of the treeview height/width based on that. The estimated non-visible size of the treeview is really only useful for driving the adjustment page size and current location used for the scrollbars. In terms of responding to height-for-width requests, it's only really important to report true height-for-width until you reach gdk_screen_height(), and only in order to support treeviews which are not placed into scrolled windows. For treeviews placed into scrolled windows (if it were the only use case), the treeview could simply report a height request of 1 pixel and not calculate anything synchronously at all. Thankyou for looking into GtkTreeView performance ;-) Kind Regards, -Tristan > > -- John > > On 12/16/2011 12:24 AM, Steve . wrote: >> John, >> >> Time to pop the window and populate. The cpu use meter (kinda real >> time, 500ms sample) shows between 500ms ans 1 second. >> >> Do you have a preferred timer method, if so i could throw a snip >> before gtk init, and maybe in the main loop idle? Now that i'm >> thinking about it, what is the best way to measure gtk activity? >> >> Suggestions? > _______________________________________________ > gtk-app-devel-list mailing list > gtk-app-devel-list@gnome.org > http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list _______________________________________________ gtk-app-devel-list mailing list gtk-app-devel-list@gnome.org http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list