On Fri, Oct 26, 2007 at 10:40:12PM -0600, Eric W. Biederman wrote:
> Adrian Bunk <bunk@kernel.org> writes:
> 
> > There isn't any hard semantics behind what is marked EXPERIMENTAL and 
> > what not. In it's current state, we could even consider removing the 
> > EXPERIMENTAL option and all dependencies on EXPERIMENTAL.
> 
> Well I do know at least some of the things that depend on experimental
> are legitimate.
> 
> I wonder if the problem is that we don't police experimental well
> enough.
> 
> > Currently CONFIG_EXPERIMENTAL=n would cost a distribution a three digit 
> > number of device drivers plus several features like e.g. NFSv4.
> 
> I can see a distribution carefully cherry picking things, that the
> have an intimate knowledge about out of experimental but it doesn't
> sound right for taking things out of EXPERIMENTAL to be routine.
> 
> I know I'm a little slow about getting around to it but when ever I
> have a feature that isn't EXPERIMENTAL anymore I remove the tag.

Part of the picture might be that code that was included into the kernel 
usually is in a state that it works at least most time for most of the 
people.

And when you think about distributions, it's hard to imagine why a 
distribution should not enable more or less all EXPERIMENTAL device 
drivers - an EXPERIMENTAL driver is much better than no driver for this 
hardware at all.

> Eric

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

_______________________________________________
Containers mailing list
Containers@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers