Re: gdk_rgb_init calls for robustness

[Havoc Pennington <hp redhat com>]

Darin Adler <darin eazel com> writes: 
> If we want to go the other way and reaffirm that we require explicit
> gdk_rgb_init calls, an alternative would be to put in return_if_fail calls
> that check if image_info is NULL in these same places, to help people notice
> when they forget to call gdk_rgb_init.

I would consider that correct - we do require gdk_rgb_init() calls for
1.2.x where x <= 8, so if we fix it in x = 9 people can accidentally
ship apps that require 1.2.9, which is not really encouraged.

In 2.0, gdk_rgb_init() is a no-op so pain here is short-lived.

Havoc