Codecov Report

❌ Patch coverage is 88.88889% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 72.76%. Comparing base (d41f87b) to head (e6497a5).

@@            Coverage Diff             @@
##             main    #2077      +/-   ##
==========================================
+ Coverage   72.72%   72.76%   +0.03%     
==========================================
  Files         635      635              
  Lines       30372    30383      +11     
  Branches     3279     3281       +2     
==========================================
+ Hits        22089    22107      +18     
+ Misses       6387     6371      -16     
- Partials     1896     1905       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

I briefly tested this out with our stack but ran into infinite loops when repeatedly re-enqueuing with a rel_time of 0 from within the action. So these new semantics introduce a case that the client has to guard against that they didn't have to before.

@tobim thanks for testing! What function are calling, exactly? delay_for/delay_for_fn? They go though the actor clock and ultimately push the action to the mailbox, so I think I need a bit more context to understand how this change could cause a regression.

Just always re-scheduling a new action from an action will be an infinite chain, so how does your workflow look like? Just from a high level, so that I can come up with a regression test.

@tobim thanks for testing! What function are calling, exactly? delay_for/delay_for_fn? They go though the actor clock and ultimately push the action to the mailbox, so I think I need a bit more context to understand how this change could cause a regression.

We are using delay_until_fn. It some situations it re-enqueues itself with the same abs_time indefinitely.

The code is at https://github.com/tenzir/tenzir/blob/65b7c9eacc1e87f9f1d6c569db29826e2b86dec9/libtenzir/src/execution_node.cpp#L45-L52.

@tobim I've made a small tweak, can you re-run your tests with that version and see if it maybe changes something for the better?