Re: GOOPS slowness [was: Compiling guile-gobject experiences]

[Andy Wingo]

On Mon, 15 Sep 2003, Andreas Rottmann wrote:

> [Cc'ed to guile-devel since this is a guile/GOOPS speed issue;

For the moment I'm just replying to guile-gtk...

>  Context: The new guile-gobject Guile bindings for GTK+ use GOOPS to
>  provide an OO interface. Building this interface (at binding module
>  load time) is painfully slow (leading to a 10 seconds total for
>  loading the whole GTK+ binding on my machine (Athlon 900)).

Your work certainly did a lot, although I can't quantify right now. What
I can say is that on my Celeron 600, it now takes 10 seconds to load up
Gtk. 10.8 according to bash's time function. Which makes me wonder about
your Athlon, which should be much faster. 

It's hard to tell, though; I was also doing a bit of hacking on the
binding generator. Really, I just wanted to clean up the informational
output when the bindings are being made, and I wanted to fix the
de-StudlyCapsification so it would get gtk-im-context and friends
properly. So I don't know if I did anything, but it's a lot less slow
than it used to be. I'll commit what I have, and we'll see if there's
any difference for you.

One note with what I have is that although it gets
GtkIMContext->gtk_im_context_* properly, the Gtk+ API is strange in that
GtkVBox->gtk_vbox_* whereas it should be gtk_v_box_*. Which way is
correct? For now, I just fixed everything to <gtk-v-box> and
<gtk-h-box>, although I don't really care either way. Opinions?

Regards,

wingo.