Respect ANDROID_AVD_HOME environment variable

[lvanengelen]

• This is a small change

---

Description:

Android SDK tools respect the ANDROID_AVD_HOME environment variable to locate installed emulator instances. This change adds support for that setting when fixing up the config.ini file.

[noomorph]

Could you check this? #1573 (comment)

[lvanengelen]

Looked at #1573 and I investigated a bit. I ran emulator -list-avds in different configurations and got the following to list the avds correctly:

• With avds inside /tmp/foo/.android/avd: ANDROID_SDK_HOME=/tmp/foo emulator -list-avds
• With avds inside /home/len/android/avd: ANDROID_EMULATOR_HOME=/home/len/android emulator -list-avds
• With avds inside /home/len/android/avd: ANDROID_AVD_HOME=/home/len/android/avd emulator -list-avds

This corresponds with the emulator source code.

I will update the branch to reflect this.

Reading #1573 I am not sure why ANDROID_SDK_HOME did not work.

[lvanengelen]

Hmm, I have a hard time deciphering the Jenkins logs to figure out what went wrong during the checks of the second commit. I also fail to understand why it would break while the first commit succeeded. Any pointers are welcome.

[noomorph]

Why not use path.join to be consistent for Windows?

[lvanengelen]

Because the original code also contained /, but I can fix that. Also gives me a way of retriggering the checks :) Or is there some easier way than pushing a new commit?

[noomorph]

If nobody looks earlier, I can look on Monday.

[lvanengelen]

That would be nice. As an experiment I force-pushed my initial commit (which completed fine earlier) and the checks also failed, so something seems to have changed in the test setup after my initial commit?

For now I re-pushed my other commits for completeness.

[noomorph]

@lvanengelen, thank you for your contribution!

[lvanengelen]

Thanks for helping out! Just out of curiosity, could you explain why CI kept failing after my earlier pushes?

[noomorph]

I am not sure about everything, but, in particular, the Jenkins OSS CI had undergone some configuration changes and only the recent commits to master have fixed android-rn-59 problems and maybe something else. I have noticed that your branch has been not in sync with master for enough time to start getting some of those issues.