[PATCH] Do not send / kernel splitting / deprecation

[Werner Almesberger]

Andy Green wrote:
> Yes indeed.  Also I know this is not a popular subject but in the same
> vein looking ahead, we need a deprecation policy for product support.

I think what makes this subject unpopular is that deprecation isn't
really a concept that applies here.

"To deprecate" implies strong negative emotions about the object of
that deprecation. In the Open Source world, we tend to develop the
occasional dislike for software (particularly when we're about to
replace it with something better), but hardware, being something
that's not so easy to replace, is usually held in much higher esteem.

If you look through the Linux kernel, you will find lots and lots of
drivers and even entire platforms that are no longer produced, no
longer sold except perhaps on eBay, for which no regular commercial
support has existed for years, and for which far superior and more
modern alternatives exist. Yet this code still gets fixed when it
breaks.

So people in our world don't deprecate old hardware. They or their
companies may lose interest, or even the ability to continue support,
but there's nothing stopping someone else from taking care of the
elderly and frail, or orphaned and abandoned.

There are usually also several stages in withdrawing support. First,
there may be the point when one starts trying to make something
better (new features, enhancements). Then, there's the point where
one stops trying to keep it from getting worse (regular bug fixes).
And finally there's the point where one doesn't even talk about it,
except perhaps when swapping stories about the good old times.

> If we can make sure that we help any community effort to take over
> development on products that go out of our post-EOL support period
> (whatever that is) this doesn't have to make a big problem.  Plus the
> userspace is pretty much shared again reducing the hit.

Yes, end of support sounds much friendlier than "deprecation".
And this is indeed how support should be phased from the hands of
OpenMoko to other people. This is something that happens all the
time, in countless projects. Not always smoothly, but as long as
there's enough interest, it will happen in one way or another.

I noticed that there seems to be a lot less activity on
openmoko-kernel about GTA01 these days than there was a few weeks
ago. Does this mean that all problems are solved now and it's
perfect ? :-) Have people given up on it ? Is everyone just hiding
from the flood of GTA02 postings that have started to fill the list ?

- Werner