Re: [gtkmm] Custom TreeStore Problems

[Christof Petig <christof petig-baender de>]

Murray Cumming Comneon com schrieb:
> In this case, I am not convinced that enough research has been done to fix
> the original performance problem. People don't seem to have this problem

Since it seems to relate to a tree being accessed by a sequential 
iterator to mimic binary search (this clearly is a problem) there is no 
way around that (besides adding a different style iterator (which 
exhibits the tree nature) to GtkTreeStore - this might be an option if 
this problem was more common). [I really should take a look into the 
glib representation behind GtkTreeStore]

Another possibility is that passing _boxed_ types around too much sucked 
on my performance. [column access being slow]

> with GTK+, and Daniel didn't seem to have this problem with gtkmm in
> regexxer. If someone can show that it's a gtkmm-specific problem then I
> would like to investigate that.

perhaps the link below gives hints about that. To my knowledge the 
performance problems of this example had not been researched (it seems 
to be not pango, not treeview related (I have a patched version which 
fills first and then attaches)).

> Basically, I like to isolate problems, put them in bugzilla, and fix them.

I agree, but after three days of profiling (I used sprof on 
(glib,gtk,gtkmm,pango-ft and -pg on a static variant of gtk,*mm) without 
_any_ real results I have no idea on how to isolate the problem further.

> > But ... this does not yet explain the problem mentioned on the 
> > gtk-app-devel list (see my previous post for the source). He did only 
> > append.
> I seem to have lost that. Do you have a link to the archives?

of course (it's in my sent folder)
http://mail.gnome.org/archives/gtk-list/2003-March/msg00361.html

   Christof