Re: Backwards compatibility issues with GDK 2.4

[Mike Hearn <mike navi cx>]

On Thu, 22 Apr 2004 08:07:27 +0100, Mark McLoughlin wrote:
> 	I really think that is the only answer because changes like this are
> quite common with any library - e.g. take this one made between gtk+ 2.0
> and 2.2:

It's hardly the only answer: requiring defines to get new functionality
(or using weak symbols) is not a hard thing to ask people to do, and it
allows the developer to control exactly when they begin depending on new
functionality without requiring two installations of their libraries.
 
> 	Basically, if you don't just say "this is expected behaviour, you
> should not expect this to work" and try and work around this particular
> issue, it doesn't solve the problem in general. We will continue to see
> things like this coming up in the future.

The problem is a matter of policy, really. The fact that this sort of
thing is common and been done in the past isn't a good reason throw up our
hands and say "oh well, that's just the way it is" - in the past free
software libraries didn't make any ABI or parallel
installability guarantees either and pain for end users and developers
alike was the result.

thanks -mike