On Saturday 25 March 2006 01:33, Axel Thimm wrote:
> On Sat, Mar 25, 2006 at 12:22:52AM +0100, Hans Verkuil wrote:
<snip>
> > > I wouldn't bother, because there is only this one kernel rpm with
> > > this mismatch, this kernel is flawed in several ways like not
> > > being able to load nvidia, ati, madwifi, ndiswrapper and the
> > > like, and a replacement kernel (true 2.6.16, rpm also called that
> > > way) is already in updates-testing.
> >
> > Yes, but it is in the initial FC5 iso download, so unless that's
> > replaced many people will just pick that and install ivtv over it.
> > And I get the complaints...
>
> I hope most FC5 users will continue to use packages, so they won't
> step into this pitfall.
>
> > Not only compile problems, it also introduces a bug in ivtv-osd.c
> > (also checks against the kernel version and picks the wrong path).
>
> You mean at run-time? That requires a bit more hacking around to
> fix. In that case a special handling workaround in the sources would
> be better.
No, at compile time. But all you see is a compile warning. The API for
two calls changed in 2.6.16, and now the compiler compiles the wrong
version.
> > It's not much work, the advantage is that this 0.6 release is for
> > the 2.6.16 kernel only, so I can just remove some tests.
>
> Wouldn't 0.6.x build on a previous kernel if it had the latest
> video4linux bits externally added? E.g. maybe it's better to keep
> some tests for such cases.
I've added tests to check for older and newer kernels, so it should be
good.
>
> What I'd like to see long term is video4linux support for 2.6.9
> kernels (I'm bugging Mauro on this) and of course having the latest
> ivtv built on this, too. That way people can create mythtv boxes on
> RHEL4 and derivatives/clones which will give them a longer lifespan
> than with Fedora Core.
Would be nice.
Hans
_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel
