build E2 with eplayer3

[rantanplan]

mhh
ich hab immer noch bestimmt überhaupt keine Ahnung und will mich auch nicht da aus dem Fenster lehnen....

Wo Santa schon auf pli Diskussion hinwies, daher
https://github.com/Taapat/vuplus-ful.../oem-fulan.inc

Eigentlich steht da schon das meiste drin, wie man zum reinen eplayerImage kommt.

Grüße

[max]

ja mag sein, war halt schon immer so ;-)
was ich aber nicht verstehe,
https://github.com/Audioniek/cdk/blo...r/make.sh#L373
das der enable-buildinplayer in zeile 373 raus ist.
Neutrino nutzt nicht den eplayer3 aus apps, sondern den eplayer3 aus stb-hal und somit wird der eplayer3 aus apps umsonst mit gebaut.

[Audioniek]

@Max:

Die Zeile ist raus, weil dieses Teil der make.sh ein case [Imagetyp] ist. Die case faengt an in Zeile 316. Zeile 373 ist im Teil fuer Enigma2, Enigma2 laesst sich mit built-in nicht bauen. Umgekehrt ist bei Neutrino und Tvheadend built-in die einzige Moeglichkeit der Sinn macht, und die werden immer mit built-in gebaut wobei es bei Tvheadend eigentlich academisch ist: der Box gibt nie video wieder.

Bin gerade daran ein ueberarbeitete make.sh zu pushen, die auch den Enigma2-diff abfragt, damit die Build nie interactiv wird. Diff und revision werden via Umgebungsvariabelen und configure.ac uebertragen (be-einflusst auch enigma2-pli-nightly.mk, die Wartungsfrei wird, also diff-Pflege nur noch im make.sh). Erscheint sobald die jetzt laufende Testbuild OK ist und ich zufrieden bin.

Gruesse,

Audioniek.

[TangoCash]

@audioniek:

Schau dir mal das an:
https://github.com/TangoCash/tangos-enigma2

Das ist eine E2 Version mit eingebautem eplayer3 (wie bei neutrino)
Zusätzlich kann man dann mit entsprechendem makefile auch so wie neutrino, nicht im selben verzeichniss wie die src sind, bauen.

[max]

ok habe ich jetzt gesehen ;-)

[Audioniek]

@TangoCash:

I hope you do not mind me answering in English. I have started out to integrate your eplayer3-E2 into a test build environment based on my git. Currently I have ended up with a very bulky enigma2-pli-nightly.mk but it builds to completion. Resulting image can be flashed into a Spark7162 and starts. There is some basic work to do as the remote does not work at this point (aotom part is OK), but the start assistant shows a picture a with a Dream remote complete with arrows and a timeout counter.

The next steps I plan to undertake is get the basic eplayer3-image working in my test build environment. When that is done, I hope to determine the differences between an gstreamer+libplayer3 OpenPLi and your eplayer3-E2 variant. This will then hopefully lead to either a patch file to patch the gstreamer stuff out, and/or a more streamlined enigma2-pli-nightly.mk. Selection of which image to build is currently done using the diff-level which is in turn evaluated in enigma2-pli-nightly.mk. Along these lines I hope to reach a point you can build any current OpenPli diff for either three framework variants, but there is quite a but of work to be done to get there and may not be achievable.

In my view image size difference is considerable: gstreamer+libplayer3 is just over 60Mbyte, yours is somewhat less than 48Mbyte.

Thanks for your input. Testing, testing, testing....

Regards,

Audioniek.

[TangoCash]

I also ported the lirc handling into e2 (https://github.com/TangoCash/tangos-...af2c3f0e51ccc2) ..otherwise my rc wont work also

I just build Neutrino/E2 Dual images, and they are about 42mb

[Audioniek]

Meanwhile I have got a running image on a spark7162 with the remote control working. Fixing the remote was relatively easy: /dev/input/nevis_ir does not exist in my standard build. Replacing it with "/dev/input/event%d" and using the standard OpenPLi code fixed it for me.

I am now capable of determining the differences between a gstreamer based E2 and yours. Most of of what I saw so far is in Makefiles and can very likely be resolved in the enigma2 diffs so E2 will build with any frame work. This is because you can test with ENABLE_MEDIAFWGSTREAMER in both Makefiles and code. I am not there yet, but getting closer to home.

The main difference is the libeplayer3 code (by schischu, konfetti and others) in the E2 lib/libplayer3 directory, which is not part of standard OpenPLi. So far I use your git to get this part from the net. I was relieved by the fact that the rest of the build environment does not need changes when building with eplayer3 only, all changes are within E2.

Patching on...

Regards,

Audioniek.

[TangoCash]

I try to keep the changes in the build environment as small as possible, and try not to break other builds, like gst, or other E2 gits in this case.

I also ported E2 to libsigc++-2.0 api, because Neutrino relies on it. (not in git, just by a replace shell script)

The eplayer3 implementation ist not finished yet. At this time the AudioPids and the SubtitlePids are just ignored

And i deactivated the servicemp3(gst based) because eplayer can handle this within. Both needs further investigation/tests.

all other modifications are to keep e2 small, and fast, or fetch my needs.

[Audioniek]

For those want to build E2 with eplayer3 only, you can now have a try. I have pushed the necessary changes to the git. All diff-levels build to completion with eplayer3 and have been tested on a spark7162. Tested in this case means: Image starts, displays and records an FTA transmitter and can play the recording back. I have not tested anything else, hence the term experimental behind the choice in make.sh.

Almost alll changes have been incorporated into the E2-diff files. The changes I could not handle this way are done in a small extra patch file.

Vielen Dank an TangoCash.

Regards,

Audioniek.