1) Hardware constraints. 2) Software must be real-time.
Non-embedded developers are usually unable to understand the constrains of embedded and chalk it up to "you are dinosaurs". Not true.
The lack of Py3 support is not about Python vs $language on $environment, it's about using EOL software in general.
[1] - my last public embedded software contribution was in fact an ELF loader for a smartwatch so that I wouldn't have to use micropython: https://git.card10.badge.events.ccc.de/card10/firmware/-/tre...
But again, I'm not a developer and I'm just getting my feet wet so my opinion is likely wrong.
Maybe they don't understand automotive is old. There is more tech debt than you can imagine.
If you're never paying back your tech debt, it's not tech debt. It's just poor software engineering practices.
Note: Python 2.7.3 or greater excluding Python 3.x, which is not supported.
I'm not a developer by trade or talent, but I'm learning and I'm starting with Python. Everything I've read and been told about Python development says to start with 3.x because the 2.x branch is officially deprecated and should only be targeted if one is intending to learn specifically for maintaining a legacy project that still uses it. I'm learning to learn, not to work, so I think I'll wait until AGL is brought up to a current version of the language before I dive in and get myself confused.
Beyond that though, I do like the premise of having the infotainment stack open source from the ground up. It would be nice if manufacturers would pay attention to this project and write open source drivers for their hardware.