FEO Weekly 2025-11-25

[armin-acn]

Participants

@FScholPer
Jan Staschulat
Shourouk Ramzy
Ulrich Freund
Pawel Rutka

Discussion

• How to proceed towards 0.5-beta release (planned for before Xmas)?

  • Would be nice to get a bigger example application than the current "mini-adas" and "cycle-bench"
  • BMW promised to work on a visualization application, but there has not been any update recently
  • It may make sense to wait until the new Rust API to com is available before spending efforts on bigger applications

• Rust API for mw::com (info from Pawel):

  • The design of the new Rust API is progressing well
  • Qorix has an implementation of the API based on iceroryx2, it is open sourced by Qorix, but not part of S-CORE
  • The Qorix implementation could be used for porting FEO to the new COM api. Ideally only the crate would have to be changed, once mw::com is ready

• Artifacts for beta release:

  • @armin-acn and @johannes-esr to clarify with @masc2023 internally the current status and what could be delivered on top of the alpha-release.
  • Code coverage: Can we already provide some tests with code coverage?

• Meeting structure FEO/DAAL/etc:

  • It has been agreed that SCORE will support multiple application frameworks, in particular DAAL.
  • It may make sense to have a single weekly meeting to cover all of these frameworks instead of a separate one for each of them. @armin-acn will clarify that with @johannes-esr.

• Dependency graph for FEO:

  • Currently FEO dependency graphs are created manually. Mercedes proposes a yaml-based IDL that would allow to also auto-generate such a dependency graph for a given system. @FScholPer proposes that FEO could make use of that.

• Integration with Orchestrator:

  • @FScholPer points that he receives questions asking why FEO and Orchestration are completely separate.
  • @armin-acn explains that the development of FEO was started after the first draft requirements had been set up at a very early stage of the S-CORE project. The first Rust version of Orchestration has been developed in parallel at about the same time. Both approaches have their advantages and disadvantages. One of the advantages of FEO is that it does not depend on an async runtime and thus might be easier to adapt to smaller platforms. And so far, there was no need to combine both features.