...
From EG discussions in 2021 it seems clear that some degree of enabling application sandboxing in a replacement demo application framework is desirable. One proposed scheme for this was attempting to adapt FlatPak adapt Flatpak into AGL, which has some appeal given it is achieving a degree of traction in the desktop Linux distribution space. However, investigation into this approach seems to indicate that the research and development effort required is perhaps beyond the level of commitment that AGL is able to invest in. Given that, leveraging systemd's sandboxing controls somewhat in the fashion of the previous application framework seems a lot more feasible, as does attempting some extension beyond that with some of the options available in current systemd. The additional benefit to such an approach would be potential synergy with Toyota's announced desire to work with upstream AGL on integrating a systemd based replacement for the resource/task management scheme in their base system for the Production Readiness EG.
...
On the effort side, a lot of initial focus has been on attempting re-use of available FOSS services and their IPC mechanisms. In practice, that has led to using D-Bus for e.g. the new application launcher, when and that has perhaps proven to be less than ideal for consumption in the web applications and with respect to potential for sandboxing application access. It seems clear that there are several forms of service such as audio mixer, radio, vehicle signaling, etc. that AGL likely needs to do some development on to enable member demos and serve as a potential starting point for member interest in upstream development. From discussion in the 2021 workshops and EG meetings into 2022, it seems that there is a rough consensus that grpc (grpc.io) seems a reasonable framework to use as a basis for such service development. A non-exhaustive list of the rationale for this is as follows:
...