> The question to keep asking is "If its not upstream, there's a reason. > If that reason is code-quality, why is it good enough for Fedora?" .... > New code, be it drivers/filesystems whatever has to go upstream, > or at the least show signs of 'going upstream soon'. ... Totally agree. In fact, the driver that I started this thread about satisfies the above - it's pretty much a given that it'll be in one of the upcoming kernel releases. On a side note, as completely new to Fedora I was very satisfied with its driver support - besides raid controller, not a hinch with any of the drivers for my hardware that came with FC4 out of the box - keep up the good work! Thanks, Tim On 5/23/05, Dave Jones <davej@xxxxxxxxxx> wrote: > On Mon, May 23, 2005 at 01:57:38PM -0400, Tim Taranov wrote: > > I would have excluded device drivers from this category or goal, > > _especially_ if they are shown to work ok under mainline kernel > > releases as well as are part of the upcoming/future kernel releases. > > The problem is a lot of drivers that live outside the kernel.org tree > are _awful_. Its not until they get submitted for inclusion upstream > that they go through peer-review and get various issues worked out. > > The question to keep asking is "If its not upstream, there's a reason. > If that reason is code-quality, why is it good enough for Fedora?" > > Another issue is the burden it adds to the distro kernel maintainer > (ie, me right now). There are two 'add-on' drivers in the Fedora > kernel tree right now (ipw2100 and ipw2200), and it's a complete > pain for me to have to run off and download extra drivers, bend > them to fit the fedora kernel, fix up any additional problems > that get reported etc. I really regret my decision of merging them, > though I know there are countless users of those drivers who are > glad that I merged them. > > New code, be it drivers/filesystems whatever has to go upstream, > or at the least show signs of 'going upstream soon'. > > Dave > >