> Hey, that's a great idea. I just ran it. Looks good.
>
> Could you add some "padding" (controls or loops), so we can see how it
> acts with a larger app?
> With larger apps the current splash method requires a few seconds to
> show on a slow machine, so your method would be great...if it doesn't
> cause side-effects we haven't noticed yet. But, Mattia will probably
> be able to comment on that.
Hmm, how much time does it save on the low end machines you have?
Mine is P200/64MB/Linux/wxGTK ( the XServer is running on another machine ).
Normal app; results from "time perl -Mblib splash.pl", putting an exit 0 just
after the Wx::SplashScreen->new;
1.890 seconds
after mutilating Wx.pm so it loads much less than Graciliano's one
( it is not fully functional, I just wanted to know the max time saving I
could obtain: it just loads Wx::App, Wx::Bitmap, Wx::SplashScreen, calls
_boot_GDI and _boot_Constant )
1.290 seconds
gain: ~600 msec
so we will be gaining less than 33% ( notice that in a real scenario I'll
need to load at least ( part of ) Wx::Event -> less time saving ).
Is that worth the additional code complication? ( <- this is not
rhetorical, BTW...)
Puzzled
Mattia
P.S.: splash.pl and Wx.pm attached for reference; some trivial modification
to Bitmap.pm and SplashScreen.pm are needed for that to work
|