Communication CFT Weekly [2025-11-10]

[LittleHuba]

Participants

@LittleHuba
@darkwisebear
@JochenSatETAS
@kalu-an
@SebastianGrob
@alexandruiulian10
@ZoranCutura
@lurtz
@mina-hamdi
@pawelrutkaq
@NEOatNHNG
@piotrchodorowski
@FScholPer
@badiienouri1987
@Abhishek2581
@sankurm
@dasoho

Agenda

1. SOME/IP-Gateway (Pawel Rutka)
2. mw::com in the GW. Should this be done to encapsulate the SOME/IP-Stack or does the lead to further resource waste (Jochen Speck)
3. Further discussion of the method concept (Jochen Speck)
4. Pull request states (@FScholPer)
5. Platform Autosd Relationship (@FScholPer)
6. Compiler flags (@FScholPer)

Meeting Minutes

1. SOME/IP-Gateway (Pawel Rutka)

• [Contribution Request] SOME/IP Gateway #1769
• All three contestants are more or less on par
• Qorix written in Rust
• 5+ people raised their hands to support Rust development for the Gateway (if we choose to do this in Rust)

Open questions:

• Is there a SOME/IP stack written in rust?

  • ETAS stack is written in C++
  • vSOME/IP is written in C++
  • @pawelrutkaq mentioned that there might be a Rust SOME/IP stack contributed to Eclipse foundation within a year

• Should we split the gateway along the ASIL boundary also by language? SOME/IP stack in C++, Gateway logic in Rust?

  • We should define the IPC interface between Gateway logic and SOME/IP stack

• Is there a performance issue if we implement the gateway in Rust?

  • More dynamic allocations are necessary during setup of skeleton/proxy in Rust
  • Hot path should have the same performance characteristics
  • Proposal: We should investigate in performance of the Rust binding base on the benchmarks here:
    https://github.com/eclipse-score/communication/tree/main/score/mw/com/performance_benchmarks
    Focus on hot-path API for now

• Is Rust ready for safety critical contributions?

  • This is a project risk and must be solved by the project
  • We should support in solving this but we can't drive this
  • Participants: @darkwisebear,

• Architecture decision requires an inspection and comparison of all three contestants

  • Focus should be on:
    • What does each proposal solve the best?
    • Find an architecture that incorporates the different ideas to a complete solution

• Proposal:

  • We kick off based on the Qorix PoC (since it is implemented in Rust)
  • Other PoCs are incorporated based on the architecture proposal
  • Architecture proposal is presented in weekly in 2 weeks (24.11.2025)
  • Participants: @lurtz (admin), @badiienouri1987, @NEOatNHNG @mina-hamdi, @pawelrutkaq, @LittleHuba
  • @lurtz will invite to meetings

2. mw::com in the GW. Should this be done to encapsulate the SOME/IP-Stack or does the lead to further resource waste (Jochen Speck)

Postponed to next weekly

3. Further discussion of the method concept (Jochen Speck)

@LittleHuba will send some meeting options

4. Pull request states (@FScholPer)

Question was in regard to v0.5 release and what changes/fixes we should still bring in
A: To the knowledge of @LittleHuba all necessary fixes are included in the v0.5 release, with the potential exception of AutoSD support.
Based on Feedback from RedHat last-minute changes may be added. Since nobody from RedHat was in the meeting, this could not be further clarified.

5. Platform Autosd Relationship (@FScholPer)

Postponed to next weekly

6. Compiler flags (@FScholPer)

Postponed to next weekly