Writing synchronisation within a Multinode job is difficult and error-prone. LAVA should assist in the following ways:
0: Recommend (later mandate) that all synchronisation calls in refactored test submissions using the Multinode Protocol are completely contained within inline test definitions. (This does prevent the use of synchronisation calls inside custom scripts but the docs need to explain the benefits).
1: Parse the combined inline test definitions and the protocol usage within the submission - without reference to any external test definition repositories - to generate a flow table. One column per distinct role. Summarise the actions within the job and ensure internal consistency. Empty rows indicate that this role is waiting for another role. Refer to external repositories simply via URL, showing which steps wait for which elements.
2: Document how this helps test writers by separating the control flow from the test flow.