[Glade-users] why using libglade is better...

[damon karuna uklinux net (Damon Chaplin)]

On Mon, 2004-10-18 at 19:23, Christopher Cramer wrote:
> On Mon, Oct 18, 2004 at 01:29:12PM -0400, Tristan Van Berkom wrote:
> > If you want to re-create the popups every time you use them then you'll
> > need to seperate them into seperate glade files, this way you will gain
> > on RAM but loose on CPU (because your program will have to reload the
> > glade file every time for your popup).
>
> You don't actually need to put them into a separate file to be able to
> recreate them. What I do is put them all in the same file, and only load
> the widget (the main app window, a dialog window, etc.) that I need,
> using the root argument to glade_xml_new().
>
> The documentation for glade_xml_new() says, "Note also that the XML
> parse tree is cached to speed up creating another GladeXML object for
> the same file", so it shouldn't actually have to reload the glade file
> for every popup. It certainly seems to me that the startup (when it loads
> the main app window from the glade file) is quite a bit slower than the
> popup creation, although I haven't profiled it.

It says that, but I can't see the code which reuses the loaded data
(i.e. the GladeInterface). It just looks like wasted memory at the
moment.

So for people using libglade I'd suggest:

 1) Use a separate XML file for each window/dialog. Otherwise when you
    call glade_xml_new() on a file it parses the whole file. Also if
    you don't pass a root element to create it creates one of each
    window/dialog, which is also wasteful.

 2) Don't keep GladeXML objects around. Load the Glade XML file, get
    pointers to the widgets you need, then unref the GladeXML object.


Damon