Base O/S upgrade

Hi,
Is there any news on next release and which O/S version it will be based on and when it will be available, the roadmap seems to be a long way out of date.

I seem to recall a thread which was discussing a move to using Centos.

Stv T

I believe we're waiting for

I believe we're waiting for Centos 6 to get released

I sure hope someone plans on

I sure hope someone plans on making very modern kernels for Centos with DVB hardware in mind. The stock Centos kernels are usually way too far behind the hardware that we use. With multimedia hardware newer is often better. Having said that, Fedora is way too volatile and leaves MythTV packagers trying to support many versions of Fedora at once. Often three or four versions of Fedora come and go during one MythDora cycle.

CentOS 6 is roughly fedora

CentOS 6 is roughly fedora 12-ish kernel base. Which should mean it'll support everything F12 does. We'll make adjustments as necessary.

Pisani

That kernel is still too old

That kernel is still too old for some of my hardware, namely a digital only soundcard with no mixer that I haven't quite been able to use yet nor the HDMI audio built into my Nvidia 460GTS (I had to install drivers from Nvidia as the 195 drivers were way too old to recognize it for even video output). Since this is a multimedia distro, I see nothing wrong with making your own modern kernel as long as you don't break any feature of the base unless it's SElinux. The other option is to backport newer versions of Alsa and DVB drivers.

hmmm, He has a point there.

hmmm,
He has a point there. Of course we all appreciate the work you guys put into to maintaining this awesome distro, but is it really the easiest thing for you do? I mean going with centos when we have a quadrillion different rigs out here in userland, are you sure its the best path? I don't have any better ideas, but if I can help spark discussion and qualm fears that aint bad is it?

Quote: I mean going with

Quote:

I mean going with centos when we have a quadrillion different rigs out here in userland, are you sure its the best path?

I'm not really sure where the concern is here. At a minimum a CentOS 6 based release would at least support what was supported in MythDora 12.

From a kernel standpoint, both CentOS & Fedora are built to support a wide variety of hardware configurations.

Plus, it may end up that we build a newer kernel anyhow. Either way, having a stable baseline (glibc, gcc, etc) will help prevent the need to rebuild all types of packages with a new release.

Pisani

OK, I'm all for a less

OK, I'm all for a less volatile base OS. The rate at which Fedora changes is quite ridiculous for this type of appliance to be ran on. Having a stable base to work with and compiling custom kernels makes more sense than having to re-write MythDora and release a new distro simply because no one is packaging Myth for a version of Fedora that is barely a year old. I noticed for some time that Jarod Wilson has Myth packages for EL which I believe Centos is built from.

Yep. Jarod's actually got a

Yep. Jarod's actually got a pretty good tree built of mythtv and dependencies already. We're really just missing the base at this point.

I am a little confused why

I am a little confused why you don't want to go with the next version of Fedora? Is it just that the updates are too frequent?

Shredder

There are a number of reasons

There are a number of reasons really.

1) The release cycle is very quick. Which is why we've never accomplished a back-to-back release. Plus there is little reason to release on every version.

2) Support & packages go EOL quickly. Since most releases are only live for 18 months, the 3rd party repos tend to go empty after that time. I personally am still using Fedora 10 as my master, but I've had to limp it along with source packages to get to 0.24.

3) Longevity. For many, when something works they don't want to really touch it; except maybe to add a new version of mythtv. With 3 or 4 year life cycles of CentOS you'd be able to do just that. You'd still get your errata, etc, and every 6 months or so a new version of mythtv would become available. Then in the interim we'd release partial / updated builds. The main thing is that we'd have a stable base to build off of.

With only 2 of us in the project, and limited availability it's a big undertaking to build from the ground up every release. Jarod's schedule was quite busy last time and so it ended up being me putting most of the package set together. Which quite frankly is exhausting. I spend the better part of my free time over 3 months to get 12.22/23 out (when we switched gears).

4) Stability. CentOS/RHEL should be pretty stable in general. One issue we've run into time and time again is not so good kernel bases in Fedora. It's a cutting edge release, and it can definitely cause headaches.

To be honest, it's becoming so easy to setup mythtv now that we've wondered if there is a real good reason to keep it around and not just make it an add on thing that you just "yum install"

Pisani

rpm

pisani wrote:

To be honest, it's becoming so easy to setup mythtv now that we've wondered if there is a real good reason to keep it around and not just make it an add on thing that you just "yum install"

Pisani

I assume that doesn't include all the remote configs, recording devices, db backup scripts, k9copy (ok that's a yum install), nuvexport, all the video extras, all the plugins, essentially http://www.mythdora.com/?q=features that are all ready to go that we all love dearly?

Well actually, all of those

Well actually, all of those things are rpm included. So we could deliver about the same experience either way.

Putting it all together in the installer just makes it a bit cleaner/easier for thos that have never done it before.

Pisani

yum groupinstall Mythdora?

Are things well-enough compartmentalized that we could groupinstall Mythdora on top of Centos 6? It would be nice being able to have the options of a stock CentOS or a dedicated MythDora install.

Jarod Recommended upgrading

Jarod Recommended upgrading the md12 kernel, do you think the new "mythdora" will already include these fixes?

http://www.gossamer-threads.com/lists/mythtv/users/473852?search_string=How%20can%20I%20debug%20these%20remote%20issues;#473852

Its been so long since I've needed to compile my own kernel........

Anthony

There is no reason why a

There is no reason why a custom 2.6.39 kernel can't be built or the 3.0 by the time this happens. As it is in MD12, I have had to upgrade the Alsa drivers (by building them against the kernel source) to recognize the Intel HDA built into my Nvidia 460GTX but the good news is that it passes 44.1, 48, 88.2, 176.4 and 192 in pcm and most variations of surround straight out HDMI. Myth's general setup shows it as capable of 7.1 channels. It actually shows that my Rotel processor is capable of 320khz sampling. Of course I had to move to the nvidia 265 drivers since 195 is not even on the radar of recent Nvidia hardware. DVB drivers could stand to be updated as well.

To answer your question Ryan, sure you can stick to the 2.6.32 kernel as long as you upgrade Alsa, DVB and many other drivers around it. For this application, a specialist kernel makes more sense than trying to make the stock one work in a multimedia appliance versus as server. Besides that, XFS filesystem performance has made a huge leap in 2.6.39 and many will want that.

HD-PVR IR fixed?

That reminds me, are the kernel HD-PVR IR fixes available and how do I upgrade to get that working?

http://www.gossamer-threads.com/lists/mythtv/users/445207?search_string=hd-pvr%20remote