[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: rawhide report: 20060103 changes



On Tue, Jan 03, 2006 at 12:17:52PM -0800, Pete Zaitcev wrote:

 > > softmac will win in the end. Heck even the broadcom driver is ported to
 > > bscape too!
 > 
 > Yes, there are partisans on every side, and we throw our weight behind
 > softmac for the time being.

Indeed, but it shouldn't be seen as an endorsement one way or the other.

 > I hope that it was thought out well between DaveJ, dwmw2, and jgarzik,
 > because it may be a mistake, like you said. The question is, how probable.
 > I have to say, I am not familiar with the code yet enough to know how
 > different softmac and devicescape are.

What's in rawhide right now may not necessarily end up in FC5.
It could be whats there, it could devicescape, it could something totally
different, it could be nothing.  This was done mostly as an experiment
to see just how well that driver works (regardless of the MAC layer)
to provide something useful for the developers to beat it into shape
for upstream submission.

As for why this one, and no others.. Simply a manpower thing.
Once this is upstream, maybe we'll tackle adm100, or ra2500
or something else.  Remember, that Fedora is a test vehicle for RHEL.
By the time RHEL5 ships, I'd really like to see a lot more
wireless cards 'just work' out of the box with no screwing around
building bits and pieces, and definitly not having to resort
to ndiswrapper & friends.  We're not going to get there overnight
however, so we're tackling this one step at a time.

I want to stress that I absolutely do *not* want to end up in a
situation like the ubuntu people have with a half dozen half-baked
random wireless drivers just thrown into the tree and forgotten about.

I've written many times here and in other places how much I regretted
merging the ipw drivers into the Fedora tree before they went upstream.
It was a real pain in the ass to rebase to newer releases, and even moreso
when the older version got merged upstream. This time around, with the
lessons learned from that fiasco, I'm hoping things work out a little
smoother.

Finally, to head off the cries of hypocrisy over our 'upstream first' policy,
yes, I agree it may seem we've one rule for one driver and one for another,
but it comes back to that manpower thing. We can carry a few drivers,
especially if we know we'll be shot of them when they get upstream soon(ish),
but accepting stuff carte blanche isn't going to happen.  We continue to
chant that 'get it upstream' mantra, but sometimes, we'll be giving stuff
a helping hand if it's something we're hacking on too, and we want to
give it a little more exposure. (Especially if its something like this,
which is a pain to get working).

		Dave


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]