On Wed, Nov 09, 2005 at 08:43:01AM -0500, Michael Krufky wrote: > Axel Thimm wrote: > >On Wed, Nov 09, 2005 at 08:07:49AM -0500, Michael Krufky wrote: > >>Axel Thimm wrote: > >>>On Tue, Nov 08, 2005 at 07:06:17PM +0100, Hans Verkuil wrote: > >>>>Once the audio/video drivers are moved to v4l then they are > >>>>promptly removed from ivtv 0.5.0. > >>>> > >>>I just checked v4l and it only build for kernels >= 2.6.12. Or do > >>>we want to have that as a minimal kernel for ivtv, too? > >>> > >>Where did you see this? If you are referring to the rule inside > >>Make.config, that instructs the compiler to only build DVB support > >>for 2.6.12 and up. This should not affect ivtv. > >> > >>We try to always keep v4l-kernel cvs backwards compatable. > >> > >How far back? > > > We *should* support all vanilla 2.6.X kernels, and also 2.4.X > kernels, but I do not know how far back 2.4.X is supported, as I > haven't yet tried it myself. > > >>If you find that v4l-kernel is not backwards compatable with your > >>vanilla kernel, please let us know on the v4l mailing list, > >> > >Already posted there. > > > >>or better yet, send us a patch. > >> > It might be helpful for me to point out the following: > > We are compatible with Vanilla 2.6.X kernels, through the use of #IF > KERNEL_VERSION and compat.h magic.... HOWEVER, some non-standard > kernels have backported newer features into older kernels, and our > tests based on KERNEL_VERSION fail in these cases. Because of this, > sometimes builds will fail on these non-standard kernels.
No, as pointed out on the v4l list, this has nothing to do with non-standard kernels. We should keep the discussion about specific bugs in v4l code on that list. What *is* relevant to ivtv is the dependency to the state of v4l cvs. The current model w/o keeping copies of migrated drivers in ivtv means that whenever v4l cvs is in a bad state like it currently is users will not be able to get ivtv code running. No offence to v4l code! Every code has to go through development and its cvs/svn/etc developement area will be bleeding, same goes for ivtv. But having this dependency means that ivtv would be able to release a stable 0.6.x only if there will be a stable v4l release containg the required modules, too. Or otherwise ivtv 0.6.x will build on Mondays and fail on Tuesdays depending on the state of v4l cvs. Hans, please reconsider keeping a copy in ivtv. Since you are currently maintaining both copies they shouldn't diverge anyway. And the default build could skip building them. I just want users/packagers to have a workaround when things break. Just to give a number, I usually roll out ivtv packages half an hour after they leave svn. Today I've used all my day on hunting v4l bugs just to see if the modules build. -- Axel.Thimm at ATrpms.net
pgpqKXXT1IR5L.pgp
Description: PGP signature
_______________________________________________ ivtv-devel mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-devel
