Thu, Jan 16
the error looks legit... might need to adjust the get_args py27 implementation?
Wed, Jan 15
flip token logic
Tue, Jan 14
This all makes sense to me.... letting @alangenfeld take a look
As discussed offline, ideally, we'd run the backfill using a dagster pipeline backfill command instead of a custom pipeline.
Mon, Jan 13
This feels much better!
Sat, Jan 11
Fri, Jan 10
add dagit settings info to dagit info panel
Thu, Jan 9
resolve conflict with run storage refactor
use distinctly constructed context object in the scheduling context
renamed, moved DagsterInstanceContext to ScheduleExecutionContext, made it schedule specific.
Ideally, we could just pass the DagsterGraphQLContext around, which would satisfy a number of interface checks (PartitionSelectionContext, SchedulingContext, etc), but it turns out to be a little awkward in Python so I punted on that to be as generic as possible.
Adds a dagit settings section to the instance config... This keeps the dagit settings as a dictionary instead of a configurable class.
that's a good question.... I always thought of dagit as being orthogonal to the dagster instance, but it makes things a lot simpler just to have this be on the instance config
Wed, Jan 8
fix missed callsite
Tue, Jan 7
we should add tests to test_run_storage.py
pull out run queueing into separate execution manager
Mon, Jan 6
Can you add https://github.com/jupyter/nbformat/issues/155 to the description?
should note, it's a breaking change that environment_dict_fn, tags_fn, and should_execute must take in the context as an arg...
Added context interface, for user-defined scheduler functions.
python 2.7 update
Sat, Jan 4
will let other folks take a look...