rawhide report: 20060103 changes

[Pete Zaitcev]

On Tue, 03 Jan 2006 09:58:10 +0100, Arjan van de Ven <arjan at fenrus.demon.nl> wrote:
> On Tue, 2006-01-03 at 03:13 -0500, Build System wrote:
> 
> > kernel-2.6.14-1.1808_FC5
> > ------------------------
> > * Mon Jan 02 2006 David Woodhouse <dwmw2 at redhat.com>
> > - Merge experimental Broadcom wireless driver

> I thought the fedora kernel policy was "upstream first" ....
> I think there's about a dozen other drivers that warrant the same
> treatment as this broadcom one... at least 4 of which are wireless
> drivers as well for which the hardware is quite popular.

You are just sore because you bought wrong card, right? :-)

Seriously, let us remember why the Upstream First policy exists.
It is there in order to prevent us from carrying drivers forever,
beholden to requests of users, unable to discard them least deafening
cries of "regression" raise to heavens.

This means that it may be permissible to have drivers which we expect
to be in the mainline soon. For example, Fedora (Rawhide) has libusual
enabled. To date, it's the only distro which does so, to the best of
my knowledge (unless we count Gentoo, where users can enable it if
they install -mm). In the same way, Fedora ships patches for ub, which
are not in mainline ... yet.

The Broadcom is considered a shoe-in for upstream because it uses
the softmac. And who else does? I am tempted to call you on your
rhetoric of 4 other drivers. Show them to me, I'm curious. Well?
As far as I am aware, you're just blowing smoke.

Right this time, I am porting the driver for prism54 based USB
adapters from Madwifi to softmac. If that works, I'll ask it to
be included into mainline. There may even be a brief period when
it may appear in Fedora while not in mainline (though not necesserily:
depends on the amount of testing it gets in -mm or elsewhere).

Naturally, if you are prescient and know that Devicescape will
win, then I'd love to hear it. Otherwise, I just do not see where
the other 4 may come from.

-- Pete