Tvheadend

Hi Jonas,

Many thanks for this. It worked for me.

I did encounter a couple of problems but got it working in the end.

I did have to run the command..

sudo apt install python-is-python3

..to get the compilation to work.

Did you not already have Python 3 installed. What other difficulties did you experience.

I ensured I kept things as simple and clear as possible all set out in simple copy/paste format to suit the beginner.

Many of the build related posts here are either old and/or cluttered to such an extent it’s difficult to follow them. The self build option is far better than relying on someone to maintain a repository.

Glad it helped. If you feel you have anything to add to the post fire away.

What version of Python3 did you install. Did you install Python3-requests too.

Hi Jonas,

Thanks a lot for posting these instructions.

I run Tvheadend on a Pi 4 in Ubuntu server 22.04.

I currently get a pre built Tvheadend from an unofficial PPA provided by Michael Marley which I set up using instructions here, https://tvheadend.org/projects/tvheadend/wiki/AptRepositories

What changes would I need to make to your instructions to build from source for Ubuntu server 22.04 on a Pi 4 which I believe is arm64?

Also, I don't know much about this, but can I build it using your commands on my Intel laptop running Arch/Manjaro and then transfer the .deb files to Ubuntu or do I need to run your commands on my Pi?

Cheers,

Flex

I don’t have an RPi4 that I could build test but all you need to do is change your build target. @Dave Pickles has outlined in detail here how he achieved it for a Raspbian 64 bit install. Adjust the settings in AUTOBUILD_CONFIGURE_EXTRA to the version you require to target your build at. https://tvheadend.org/boards/5/topics/43309?r=45167

Jonas Lang wrote:

What version of Python3 did you install. Did you install Python3-requests too.

I have python 3.10.6 installed. python3-requests is already installed.

I've got the 4.3 tvheadend installation working but I've got a problem with some muxes failing their scan. I'll write a separate note once I've done more investigation. I'm not sure what the problem is. Some muxes scan and identify channels correctly but others fail. There doesn't seem to be a clear pattern.

Ok. I’ll check the autobuild script on that Python issue you had. I use Python so it wouldn’t have shown up in my build process.

The failed mux scanning sounds more like a hardware issue. Is the pass/fail pattern showing between higher and lower band Muxes. Assuming you are using a universal LNB have you the correct settings in TVH to reflect this.

Martin Davies wrote:

Jonas Lang wrote:

What version of Python3 did you install. Did you install Python3-requests too.

I have python 3.10.6 installed. python3-requests is already installed.

I've got the 4.3 tvheadend installation working but I've got a problem with some muxes failing their scan. I'll write a separate note once I've done more investigation. I'm not sure what the problem is. Some muxes scan and identify channels correctly but others fail. There doesn't seem to be a clear pattern.

By any chance did you keep a copy of your build log.

Jonas Lang wrote:

I don’t have an RPi4 that I could build test but all you need to do is change your build target. @Dave Pickles has outlined in detail here how he achieved it for a Raspbian 64 bit install. Adjust the settings in AUTOBUILD_CONFIGURE_EXTRA to the version you require to target your build at. https://tvheadend.org/boards/5/topics/43309?r=45167

Thanks for this. But I'm not sure what target to put in the AUTOBUILD_CONFIGURE_EXTRA command. Here is my system:

ubuntu@ubuntu:~$ uname -a
Linux ubuntu 5.15.0-1015-raspi #17-Ubuntu SMP PREEMPT Mon Sep 12 13:14:51 UTC 2022 aarch64 aarch64 aarch64 GNU/Linux

The tvheadend git autobuild script page https://github.com/tvheadend/tvheadend/tree/master/Autobuild has some script options but none look like what I need for Ubuntu 22.04 Server on a Pi 4.

Why does yours not specify a target after ./Autobuild.sh ??

I've seen ..../Autobuild.sh -t xenial-arm64.sh in some places and @dave pickles uses

./Autobuild.sh -t raspbianbuster-armhf

It all seems very intimidating to me, I could be facing walls of cryptic errors messages for days.

Cheers,

Flex

I was talking about building on your laptop for the RPI4.

If you are building on the RPI4 you can see there are builds for the different processors. Part of the build process is a check on what processor is being used so you’ll get the relevant latest build built for your board.

My instructions should generally cover most Debian based distributions but all I can confirm working is what I built mine on.

Remember the autobuild script does all the heavy lifting and advises you if you are missing something that you can install and build again. The messages are not cryptic.

The best thing to do is go ahead and start the build process. What harm can it do and post back if you get stuck along the way.

Martin Davies wrote:

Hi Jonas,

Many thanks for this. It worked for me.

I did encounter a couple of problems but got it working in the end.

I did have to run the command..

sudo apt install python-is-python3

..to get the compilation to work.

As I suspected.The autobuild script is targeting Python to complete the build and not Python 3. python-is-python3 will create symlinks pointing the build process to Python 3 when it looks for Python. The devs may want to change the build script to reflect. I've changed my instructions above to reflect this.I ran the build again using these changes and deb package built with no errors.

Hopefully others can now use these build instructions just to confirm all is in order.

Jonas Lang wrote:

The best thing to do is go ahead and start the build process. What harm can it do and post back if you get stuck along the way.

Hi again,

OK I tried running your commands by ssh'ing in to my Ubuntu 22.04 running on a Pi 4.
When I tried the first command I got:

E: Package 'libavresample-dev' has no installation candidate

I then got to..

AUTOBUILD_CONFIGURE_EXTRA=--enable-libffmpeg_static\ --enable-trace\ --enable-debug ./Autobuild.sh

But after about 20mins got the error:

"configure: error: cannot guess build type; you must specify one" 

Just to recap, these are my system details (Ubuntu 22.04 on a Pi 4, arm64):

ubuntu@ubuntu:~/build/tvheadend/master$ uname -a
Linux ubuntu 5.15.0-1015-raspi #17-Ubuntu SMP PREEMPT Mon Sep 12 13:14:51 UTC 2022 aarch64 aarch64 aarch64 GNU/Linux
ubuntu@ubuntu:~/build/tvheadend/master$ cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=22.04
DISTRIB_CODENAME=jammy
DISTRIB_DESCRIPTION="Ubuntu 22.04.1 LTS" 

I then tried:

AUTOBUILD_CONFIGURE_EXTRA=--enable-libffmpeg_static\ --enable-trace\ --enable-debug ./Autobuild.sh -t bullseye-arm64.sh

target bullseye-arm64.sh not supported

Likewise:

target xenial-arm64.sh not supported

I edited xenial-arm64.sh and made a jammy-arm64.sh like this:

AUTOBUILD_CONFIGURE_EXTRA="${AUTOBUILD_CONFIGURE_EXTRA:-} --arch=arm64" 
DEBDIST=jammy
source Autobuild/debian.sh

I then tried:

AUTOBUILD_CONFIGURE_EXTRA=--enable-libffmpeg_static\ --enable-trace\ --enable-debug ./Autobuild.sh -t jammy-arm64.sh

AND

AUTOBUILD_CONFIGURE_EXTRA=--enable-libffmpeg_static\ --enable-trace\ --enable-debug ./Autobuild.sh -j$(nproc) -t jammy-arm64.sh

But each time I got back:

Untracked files:
  (use "git add <file>..." to include in what will be committed)
        Autobuild/jammy-arm64.sh

nothing added to commit but untracked files present (use "git add" to track)
target jammy-arm64.sh not supported

Maybe I just need to install Ubuntu on my intel laptop and run your commands from that.

Flex

As I mentioned above I don’t have an RPI to test build. I linked that Dave Pickles post to highlight what he did successfully on an RPI4. https://tvheadend.org/boards/5/topics/43309

Yes I suggest you try replicate my hardware setup and use the build instructions I outlined above. Once you succeed and get a feel for it you could probably attempt a cross compile for the RPI4.

I’m building on an i7 on Ubuntu 22.04 I’ve also built it on a Core 2 Duo successfully too.

Rather than fighting the automation it might be simpler to compile manually.

        cd /tmp
        mkdir -p makepkg
        cd makepkg
        git clone https://github.com/tvheadend/tvheadend.git
        cd tvheadend

        echo "LANGUAGES_ALL = en_GB" > Makefile.common

        ./configure \
                --prefix=/usr \
                --enable-ffmpeg_static \
                --disable-ffmpeg
        make -j3
        make install

Code compiled on an Intel laptop will definitely not run on a Pi4!

Jonas Lang wrote:

By any chance did you keep a copy of your build log.

Not sure. If you let me know where to look I'll report back.

[apologies for the delayed response - I replied by email but the response bounced.]

Dave Pickles wrote:

Rather than fighting the automation it might be simpler to compile manually.
[...]
Adjust the languages line to your needs. There are lots of options for the ./configure command to fine-tune the build; enter ./configure --help to see them all.

Code compiled on an Intel laptop will definitely not run on a Pi4!

Thanks for joining in on this Dave Pickles.

As can be seen from the title of my post this post was in response to my experience of trying to build TVH for my own i7 PC with onboard Intel graphics running on Ubuntu 22.04 64bit.

I spent a considerable amount of time trawling through numerous posts with many irrelevant comments until I found enough information to successfully compile TVH for my own setup. I decided to share my working solution here for the benefit of other beginners in the shape of simple copy/paste commands. I was hoping others would try this out and report back for the benefit of other beginners

I knew that as soon this post would be seen the immediate response would be “can I use these instructions to build TVH for my RPI4”. Again I don’t own one of these so I’ve no way of knowing.

Now that you have joined the discussion could you possibly provide a similar process for building for the RPI4. Believe me there’s plenty of demand for it. The danger of providing build instructions for the RPI4 in the body of this post is it will get completely lost and won’t show up in a search.

There’s so many new RPI4 users coming across TVH for the first time with little or no Linux experience who would appreciate some form of automated solution for building and installing TVH on an RPI4. Contrary to what I’ve read here and elsewhere TVH is far from dead.

I was replying to Sean Warner, who is trying to build Tvheadend for Ubuntu on Pi4.

The automation scripts create a package for the chosen architecture and distribution which the user can then install. Sean's problem is that there is no automated build for Ubuntu on Pi4, and trying to create one seems to lead to even more problems. Hence the suggestion to build and install directly on the target machine. I've successfully done the same for Arch Linux on Pi2.

The ideal solution would be for the project to create a new release from the current development branch. The various distributions would then take on the task of packaging it. Unfortunately there seems to be no-one with the time + inclination + knowledge to do that.

That’s exactly what I was referring to.

My instructions did not include build instructions for anything other than the x86_64 architecture. That’s what I tested those build instructions on. I’ve just completed a fresh build of TVH this evening using them. I did mention that they could possibly be amended to use for building on the RPI4 but as you’ve pointed out this does not appear to be the case.

Maybe a fresh post on the subject of building for the RPI4 using Ubuntu would attract more interest from RPI4 owners on the forum.

I’m glad to see someone like yourself is back contributing on the forum again. The forum could certainly do with your input. It was going down quicker than a lead balloon.

I actually have quite a lot of builds already going through in my fork which will "soon" be merged into master and then build the new official packages. Unfortunately some builds for old os's aren't able to upload to the repo yet due to old python but other than that it's building for more than 40 targets, including various raspberry pi os's. The fix for the python issue is the same for all those that are failing right now so once that's fixed things will move pretty fast, I just need to find some time to do it.

I for one truly appreciate the efforts yourself and others are making to continue to support TVH development across multiple platforms. It’s not easy and very time consuming to continue supporting this project on a voluntary basis. Hopefully others here will bear this in mind when commenting on your efforts. Thanks again.

Just updated my original post to include simple copy/paste instructions to keep your local version of the TVH code in line with the GitHub master once you’ve built your first version . You’ll never have to be without the latest TVH Github master commit.

Dave Pickles wrote:

Rather than fighting the automation it might be simpler to compile manually.
[...]
Adjust the languages line to your needs. There are lots of options for the ./configure command to fine-tune the build; enter ./configure --help to see them all.

Code compiled on an Intel laptop will definitely not run on a Pi4!

Dave Pickles

I used your manual build instructions to build TVH 4.3.0 on Ubuntu 18.04 on an old Core 2 Duo laptop. It went reasonably well and is up and running now.

Two questions I have. The build didn’t daemonise TVH. I had to start it from the CLI. Is there an option to include a daemonised service at build. Also does the build process you outlined have a make uninstall option available.

The build didn’t daemonise TVH. I had to start it from the CLI. Is there an option to include a daemonised service at build.

Ah, you need a systemd service file. This is the one I use:

[Unit]
Description=tvheadend
Requires=video.mount
After=network.target video.mount

[Service]
Type=forking
PIDFile=/run/tvheadend.pid
ExecStart=/usr/bin/tvheadend -f -p /run/tvheadend.pid -u hts -g video
ExecStop=/usr/bin/rm /run/tvheadend.pid
Restart=always
RestartSec=30

[Install]
WantedBy=multi-user.target

Also does the build process you outlined have a make uninstall option available.

This process completely bypasses the package management of the distribution so there isn't an uninstall option.

Thanks Dave

Yes I created a Systemd service earlier something along those lines to launch TVH and have it running as a service.

It’s definitely an informative way of building TVH and quite straightforward once you’ve had a go. Definitely gives a better idea of all the build options available. I’m going to try again just to get up to speed on the process and familiarise myself with it.

I assume you need to manually remove the install but again I’ll do a bit more reading up on the process.

Once again thanks for the pointers. It’s easy once you eventually know how.

I have tried all the guides I could find. I used different machines and they all fail when they got to this line. Does anyone know how I can fix this?

CC              src/htsmsg.o
CC              src/htsmsg_binary.o
src/htsmsg_binary.c: In function ‘htsmsg_binary_des0’:
src/htsmsg_binary.c:80:39: error: writing 1 byte into a region of size 0 [-Werror=stringop-overflow=]
   80 |       ((char *)f->_hmf_name)[namelen] = 0;
      |       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~