Challenge #3e: Efficient Broadcast, Part II

Any hints on achieving less than 20 messages per operation?

Can I get less than 20 messages without batching messages? Is there a clever topology trick?

Was considering batching broadcast updates, but I feel like I’m cheating. Peeked bit in maelstrom’s 4-crdts and Challenge#4. Seems to me like batching is supposed to be used for the next challenge since maelstrom’s 4-crdts cover g-set and g-counter and Challenge#4 mentions g-counter at the top.

Should the solution still succeed with --nemesis partition?

1 Like

Can I get less than 20 messages without batching messages? Is there a clever topology trick?

Topology adjustments work for challenge 3d but that won’t be enough for 3e.

Was considering batching broadcast updates, but I feel like I’m cheating.

Not cheating at all! You’re on the right track. There’s no one “right” way to do many of these challenges. Sometimes there’s multiple approaches and batching is a great option for some of them—3e in particular. :slight_smile:

Should the solution still succeed with --nemesis partition?

Yes, it should still work with partitions being injected. Maelstrom will remove the partitions before the end of the evaluation so your program will have a window to resolve outstanding replication.

1 Like

That worked great. Been enjoying them so far, thanks for putting this together.