Re: [wli@holomorphy.com: Re: NMU: kernel]
On Wed, May 26, 2004 at 02:29:18PM +0200, Sven Luther wrote:
> > > Any different than the ppc patch we have now ?
> >
> > It wouldn't be in the default kernel-image.
>
> Yeah, so you are going to massively duplicate the amount of kernel-image
> available. I am not entirely sure this is the wisest thing to do.
No, it shouldn't go into any shipped kernel-image.
> > > About the announcement, that would be the responsability
> > > of the author of the patch, no ?
> >
> > Usually yes. Not that there are any clear rules..
>
> hehe, but we are here to set those, no ?
Well, we don't make lkml rules here. If someone isn't the maintainer
and wants to do the "PR" work in coordination with the maintainer that
shoukd be okay.
> > Having features patches applied and not platform-depend fatures on only
> > one architecture completely defeats the idea of a portable operating
> > system
>
> Well, sure, but right now, most patches are 2.4 only anyway, so i
> believe you will have a more challenging issue with making sure the same
> patch applies to 2.4 and 2.6 anyway, don't you ?
patch don't apply to both 2.4 and 2.6 ever. Having patch that apply to
either 2.4 or 2.6 for all architectures should be the way it works. For
2.4 it's pretty hopeless, but let's get it right for 2.6.
> > Of course there's a difference, mostly due to different release cycles.
> > But that really applies to fixes only and not to new features.
>
> Or so you decreed ? What if there is something of value to our users,
> but which has not gone into the mainstream kernel, should we add it or
> have the user wait another 2+ years for this feature, which maybe will
> go into the mainstream kernel a few weeks after the new release ?
Or try to work with upstream.
> > Code review can never hurt, lack of code review hurts very often.
>
> Well, what about user-testing feedback ? Remember we are working on
> debian/unstable, which is a different situation than debian/stable.
Doesn't hurt either. And user feedback a day or two later shouldn't
be a problem in general.
> > You can test a kernel package even without an upload to the archive, and
> > if you're experiemting around that shouldn't be done anyway. Just
>
> Well, but in the case of very unresponsive userbase, the way to get
> experimenting is uploading, and wait for people cry when it breaks.
> Clearly making a call for tests has had very small feedback in the past,
> at least on powerpc. And imagining that i will have access to all kind
> of powerpc hardware and do the test myself, is clearly unrealistic.
Well, if the userbase is unresponsive they shouldn't complain to have to
wait a day or two more because of code review. It's not like there are
daily kernel releases for debian anyway.
> > So public discussion and review of patches if bureaucracy for you? I
> > tend to very strongly disagree, it's absolutely nessecary for high
> > quality software.
>
> Sure, but not at the expense of the freedom the current system gave us.
taking the freedom of an individual to put unreviewed changes into the
debian kernel away sounds like a good tradeoff to me to get better
quality releases.
Reply to: