Friedrich Beckmann

> I think the current problems I see are due to llvm updates: https://bugreports.qt.io/browse/QTBUG-104580 I think the problem is indeed a changed llvm version. The bugfix is already in QT:...

@zdanek if you look at the last CI run for Android 32bit for commit ["libevents: update submodule"](https://github.com/mavlink/qgroundcontrol/runs/8031541083?check_suite_focus=true) you will find the following error message in line 4241: `/usr/local/lib/android/sdk/ndk/25.0.8775105/toolchains/llvm/prebuilt/linux-x86_64/bin/llvm-readobj: error: unknown...

The github ubuntu-20.04 build machine provides the android sdk and ndk. See: https://github.com/actions/runner-images/blob/ubuntu20/20220821.1/images/linux/Ubuntu2004-Readme.md

I do not have an Android tablet. Can maybe somebody test the resulting apks from the pr?

@zdanek @booo : Thanks for checking the PR #10380.

@zdanek : This breaks the regression?

> Would it implicate any downside from your part regarding the work you are doing on the rest of gstreamer commits? Thank you! Hi @Davidsastresas, I think this one is...

After having digged in this dirt for quite a while my impression is that one problem might be the use of the more sophisticated "decodebin", "decodebin3" and "parsebin" technology. This...

Hi @patrickelectric, I agree the current pipeline with parsebin + decodebin(3) promises to automagically select the correct parser and the possible hardware accelerated decoder. However see: https://github.com/mavlink/qgroundcontrol/pull/10420#issuecomment-1353152604 => parsebin +...

I dived into this 100m deep tank of shit of the gstreamer video stuff some time ago: https://github.com/mavlink/qgroundcontrol/pull/10420 My result is that it is impossible to setup a automagic pipeline...