I will commandeer this (or the origin diff)

move eventlog stuff to new file please

some other options to consider:

no real cost to this though so whatevs

whats the point of slots? just speculative memory optimization?

the only thing to consider is moving some stuff out to separate files

reuse solid handle to and use that to index for compute step

just looking for answers to [1]

do we have any tests for dagre?

dagster schedule list
dagster schedule running

cleamn

so this takes care of everything so its just pip install dagster-aws; dagster-aws; (include in test plan)?

would love to see a stacked diff on top of this that takes a pass at cleaning up the weird class hierarchy / poor naming

rebase

comment

youve got enough code here to warrant breaking things out to a few files, now or in a subsequent diff

oops forgot to add reviewers

to your queue for restacking on some lower level clean ups

rebase

up

ideally we also wouldn't even watch the log files for external runs unless we had an active subscription on them

we could land this after a bit of cleanup - but i think im tempted to just have this feed in to a more systemic refactor of all this PipelineRun stuff.

open and close connections every time

defering to max

should put this somewhere other than bin?

looks pretty solid to me

hold connection

rebase

rebase

rebase

might be worth moving from init to a couple smaller files

rebase w/ fixes

use decorator

fixes

comments

rebase

fixe

add on_pipeline_init and on_pipeline_teardown methods

clean up

this looks good to me but ill give @bengotow time to look

cant remove the log_step_event call

fixes

what does dagster cli output look like on hello world?

can you throw a big comment block somewhere in here indicating exactly what this is? I could see some one hitting this too early and thinking its "the way" to do things

rm spurious changes

open connection every write

try/finally & faster poll rate

lint

fix dask

rebase

fix bad callsite