07/19/2021, 10:26 PM
ok, the above was resolved. there is one more interesting aspect which i don’t think that
can tackle, but which is probably worth looking at in a non-bugfix followup:
we choose the number of swimlanes based on the number of processes that we can run locally
but we will do many hundreds of local/remote cache lookups concurrently before we attempt to acquire the semaphore to run a local process
we don’t have a useful way to render that really wide parallelism, but it would probably be useful to.
one model that i’ve thought about in the past would be to have a line rendered in the dynamic UI containing stats for the run (local/remote cache hits, ongoing cache lookups, etc)
07/19/2021, 10:37 PM
how similar is that to John's guage suggestion?
07/19/2021, 10:38 PM
this would be a magical line that was an aggregate of the entire run, whereas John’s suggestion is that individual workunits should be able to render gauges in their descriptions