<vilhalmer>
in the build request distributor, popNextBuild bails out early if the worker it first pops is unable to provide any worker to process the build; later on it looks through all the workers
<vilhalmer>
it seems like it should either be a continue instead of a break, or just wait until it gets to looking through all workers
<vilhalmer>
right now I'm seeing only a single build request being considered for scheduling each time it wants to schedule any, because my custom nextWorker function can't find a worker for the one at the head of the queue
<vilhalmer>
am I missing something with the intention of that break?
<vilhalmer>
I think I can work around this by using canStartBuild instead to say no to each worker in turn