[lsb-discuss] X11 Extensions for Fullscreen OpenGL

Jeff Licquia jeff at licquia.org
Mon Apr 28 11:24:47 PDT 2008

Jan Emoti wrote:
> Whose demanding is important, particularly when some hardware and 
> software vendors have an incentive to discourage reverse-compatibility 
> and encourage planed-obsolesce, wasteful rewrites, and forced upgrades. 
> Those that complain the loudest are often working toward increasing the 
> short-term churn. They may have the best of intentions, just wanting to 
> be appealing or to win, but the consequence is everyone is left with a 
> less mature platform to build on.
> If the baseline to be LSB-compliant is only set in the near past, those 
> vendors above will be in control and all those millions poor users that 
> end up the world's old hardware will not be running linux!

The reason standards like Xxf86vm aren't in the LSB was to avoid 
precisely this problem: attempts to do new things often don't get it 
right the first time, and when the "done-right" implementation comes 
along, then we're stuck supporting an old technology no one wants to 

It's different when a mature technology goes obsolete (such as Qt 3), 
because everyone has a strong incentive to support older software for as 
long as possible.  But it doesn't look like Xxf86vm was in the category 
of "mature technology".

> Do unit tests for the lsb always require passing a test suite? What 
> about a library that uses a large variety of hardware where someone may 
> have to be perceiving the correct output. Does this have something to do 
> with the "lack of tests" and "completeness" mentioned on your website in 
> regard to the multimedia spec? 

We are not (yet) in the driver business, so questions of interaction 
with specific hardware aren't an issue for us.  What we care about is 
the application API: that a hardware abstraction layer, in general, acts 
as it should.

Usually, we simulate hardware in our tests, to avoid failures due to 
buggy drivers.

More information about the lsb-discuss mailing list