On Friday, September 17, 2010 12:59:09 Andy Walls wrote:
> On Fri, 2010-09-17 at 10:30 +0200, Hans Verkuil wrote:
> > On Friday, September 17, 2010 00:58:45 Andy Walls wrote:
> > > Hi Hans and Hans,
> > >
> > > I'd like to move the source code maintained here:
> > >
> > > http://ivtvdriver.org/svn/
> > >
> > > to someplace where it may be less likely to suffer bit rot.
> > > I was hoping the v4l-utils git repo would be an appropriate place.
> > >
> > > Do either of you have any opinions on this?
> >
> > I agree with this, but it would require some work.
>
> OK. But given your comments below, I'm going to change my execution
> strategy: move or port things incrementally, the most used/useful stuff
> first.
>
> Distribution packages and end users might not like that ("Where did
> ivtv-tune go?") but I don't have a solid block of time to do things all
> in one go.
Distros should always (I hope) distribute v4l-utils these days, so together
with the remaining ivtv-utils the end-user still has a full set :-)
There is nothing to be done about this, and it is quite right that you tackle
this one utility at a time.
>
>
> > xf86-video-ivtv really belongs at freedesktop.org as part of the X video
> > drivers.
> > Nobody ever made the effort to integrate it there, but it would be the right
> > thing to do.
>
> OK. That's going to be low on my list then. Given my current
> availability, that might end up looking like a "dump and run" on
> freedesktop.org folks. That wouldn't be good for anyone. I suppose I
> should get Ian Armstrong's input on this particular item.
You should certainly do that. But I don't expect this X driver to change
much. And with a bit of luck X API changes might be done for you if this
driver is on freedesktop.org.
>
>
> > ivtvtv is more a private test application I made to test the MPEG
> > encoder/decoder
> > API. Either this has to be turned into a more full-fledged application for
> > the
> > PVR-350 and then it definitely might be a good candidate for inclusion in
> > v4l-utils, or it stays here, or I can host it on my website as a simple tar
> > archive.
>
> Ok, that one will stay on ivtvdriver.org until it becomes better or
> overcome by time.
>
>
> > Looking that ivtv itself: in utils we have ivtv-radio, ivtvplay,
> > ivtv-mpegindex,
> > ivtv-tune and cx25840ctl.
> >
> > ivtv-radio is a good candidate for v4l-utils, but it would be really nice
> > if it
> > could be made more general. E.g. v4l2-radio.
>
> Yes. There is not other application like it. I also think that a
> v4l2-radio would be a good simple candidate for exercising some of the
> Media Controller API:
>
> 1. Does this device have a radio?
> 2. What is the radio control node?
> 3. Does it have an associated ALSA card? What is the ALSA PCM node?
> 4. Does it have a V4L2 PCM node?
And what would be also great is if it could handle RDS data. We still need a
good test tool for that...
>
>
> > Ditto for ivtv-tune (although merging this functionality into v4l2-ctl is
> > also
> > an interesting option).
>
> Yes. The only reason I use ivtv-tune is that v4l2-ctl requires me to
> know my analog channel frequencies by number just to change the
> channel. :(
>
> Given that, I use ivtv-tune a lot.
>
> Merging it's functionality into v4l2-ctl seems like the way to go. I'd
> like to do it in such a way that a user preferences file isn't saved
> though.
I frankly always intended to add this functionality to v4l2-ctl, but I never
got around to doing that. The frequency tables are already in v4l-utils.
>
>
> > cx25840ctl should really be integrated into v4l2-dbg.
>
> Hmmm. That would be an oddly specific addition to v4l2-dbg. Given that
> we have device nodes on subdev's coming soon, with subdev controls,
> maybe this is obsolete?
Actually, similar functionality already exists in v4l2-dbg for some devices.
The purpose of cx25840ctl is to set/get registers using symbolic names, so that
has nothing to do with subdev controls (those are definitely not meant for
debugging, instead their purpose is to provide more fine-grained control over
the subdev).
> Anyway, I never use it. Which begs the question: If I don't use it, who
> does?
I never used it either. Mainly because I find the symbolic names useless since
all datasheets are always organized around the register address and not the
name.
So if I have to use the name then I am forever searching like mad in the
datasheet. I much prefer a raw register address with a useful comment over a
symbolic name.
Looking at v4l2-dbg I would say that it should be easy to add at least the
symbolic register names to that tool. And then we can kill cx25840ctl.
>
>
> > ivtv-mpegindex isn't ivtv specific and can go to contrib/test.
> >
> > ivtvplay I'm not sure about. There is a lot of overlap between ivtvplay and
> > ivtvtv.
>
>
>
> > Regarding the test utilities: some of these can go to contrib/test, some
> > that are really ivtv specific can go to contrib/ivtv. And some should
> > perhaps
> > just die.
>
> Thanks for the insight on where to put things.
>
> I agree, some things should just die. That's part of what I meant by
> bit-rot.
>
> > Basically ivtv-utils is a bit of a mess: the really good stuff has already
> > been
> > moved to v4l-utils (v4l2-ctl and v4l2-dbg both came out of ivtv and are now
> > maintained in v4l-utils).
> >
> > So the best thing is probably to clean up the useful utilities and test
> > tools,
> > making them generic where possible, and kill off the rest.
>
> Yes. Good strategy.
>
>
> > It could be interesting to hear which utilities from ivtv people actually
> > use.
>
> I'll poll the ivtv-users list later today.
>
> My most used and useful:
>
> 1. ivtv-tune: I can't remember freq <-> channel mappings
> 2. ivtv-radio: There is no other CLI tool like it.
>
> Others I've found useful at times:
>
> 3. ps-analyzer: Our version can decode MPEG Private ivtv/cx18 VBI
> packets
Definitely useful in contrib/test.
> 4. ivtvtv: Part of it was useful to test the VIDIOC_G_ENC_INDEX ioctl()
It would be so nice if someone actually had the time to clean this one up.
It is really meant as a testbed for (mostly) the decoder part of the PVR-350.
But it could also function very well as a playback utility demonstrating the
MPEG decoding API.
Regards,
Hans
>
>
> Regards,
> Andy
>
> > Regards,
> >
> > Hans
>
>
>
--
Hans Verkuil - video4linux developer - sponsored by TANDBERG, part of Cisco
_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel