4KSTACKS again

[Jeff Spaleta]

On Tue, 13 Apr 2004 21:40:44 -0300, Ben Steeves <bcs at metacon.ca> wrote:
> They're about developers *and* community-based testers, without whom the
> developers would be severely hampered in their ability to test new
> releases.
> I've noticed a lot of vociferous people on this list *without*
> ___ at redhat.com addresses whose attitude really turn me off
> contributing.  The redhat folks usually seem much more level-headed.

Don't get ahead of yourself...
community testers are important....but too many cooks spoil the
soup..and its important for testers to come into the testing process
with some perspective and the correct expectations for the experience
to be valuable.  It is absolutely wrong to expect a piece of hardware
or a specific feature to be working during the testing process. It's
always desirable, but the realities are there are a number of
competing interests that the developers have to work with, and those
interest have to be prioritized.

Asking the developers to downgrade a feature because another buggy
piece of software isn't working right is absolutely the wrong
solution. The right solution is fixing bugs and sadly with the nvidia
driver its very difficult for anyone in the community to step up and
contribute to fixing the nvidia code associated with the driver.
Everyone's hands are tied in this specific case, but its certaintly
not my place as a tester to tie the developers hands even further on
developement because we as a community do not have the access to the
nvidia source code.  There other examples where community people are
contributing code, and helping out to fix the problems associated with
the ongoing forward progress with things like the kernel.  The
solution is not to downgrade, the solution is to fix the bugs. 4kstack
is not a bug...its a feature. It exposes other bugs in other code. The
other code needs to be fixed. There are another of small and large
issues associated with development decisions with regard to the kernel
2.6 that require other code to be fixed or written from
scratch...synaptics touchpads, the amd64 kernel port....these are
non-trivial problems that people in the community are actively
contributing to to try to solve. The solution is to fix the bugs, not
to dowgrade features to work around buggy code. As much as everyone
really really wants to have working 3d support, the harsh reality is
very few people in the community have the access and the experience to
do anything proactive to solve the problem.
It sucks, we just have to wait for nvidia to make the code changes.

Community testers are valuable for feedback, but testers need to come
into the process with the understanding that they are not setting
priorities directly and the need to be okay with the fact that they
are not setting development priorities. And specifically with regard
to nvidia,  there is a history here, the nvidia driver has broken with
kernel changes in rhl betas and it will continue to break in fedora
test releases.

-jef"second and final post in this thread, to avoid a long pointless
flamewar"spaleta