- From: Joel Becker <jlbec evilplan org>
- To: Cody Russell <bratsche gnome org>
- Cc: Havoc Pennington <hp redhat com>, Xavier Ordoquy <mcarkan users sourceforge net>, Sven Neumann <sven gimp org>, Adrian Feiguin <feiguin magnet fsu edu>, Owen Taylor <otaylor redhat com>, gtk-devel-list gnome org
- Subject: Re: GtkSheet
- Date: Wed, 13 Mar 2002 18:26:27 +0000
On Wed, Mar 13, 2002 at 12:54:33PM -0500, Cody Russell wrote:
> Alternatively, a runtime method could be used wherein widget collections
> are loaded with GModule, and gtk_init() could be told what modules to
> preload at startup to save time later on.
Try to hand-copy a GTK installation for quick testing (say, on a
machine where make is not and never will be installed). Copy bits.
Discover you've missed the pango modules. Crap. Find and copy those.
Discover you've missed the input method modules. Crap. Find and copy
those too. Discover you've missed the pixbuf loaders. Crap. Ok, go
get those as well.
GTK+ is already a large set of dependancies and files. I'm not
saying it shouldn't be as it is. There are very good reasons why pango
and input and pixbuf have sub-modules. But I don't think we should add
tons of modules where we really require them all anyway.
Consider Mozilla. It has >100 shared libraries. It loads most
of them at all times. This is a noticable impact on start time due to
all the file seeking and relocation work. Again, they have reasons, and
since I don't know them all, I'm not going to blindly debate them here.
However, we should consider these issues before GTK+ becomes
libgtkobject.so, libgtkwidget.so, libgtkwindow.so, ...
Life's Little Instruction Book #407
"Every once in a while, take the scenic route."
jlbec evilplan org
] [Thread Prev