Is there an auto formatter or format command associated with this? It would be good to add a command to the Makefile if there is.
Happy to abandon if not needed
Tue, Oct 20
Mon, Oct 19
Sun, Oct 18
Taking a closer look, the new slack client is a pretty substantial refactor so it's probably not worth doing a seven style import. I think we may just want to bite the bullet and push a major breaking change to this library. If we do so, however, users who are stuck on py2 won't be able to use slack solid hooks etc.
@prha should also take a look at this
Nice catch - this is a great fix. We should note this in the change log, because it has the potential to break some people's schedules/repositories.
Fri, Oct 16
This is a breaking change, so we should think through this. We could consider doing a seven style import based on py2 vs py3
Thu, Oct 15
I think this still needs some visual treatment but this is definitely better than before.
This is great. I think the big question here now is whether we should always use cron_partition_range and create_cron_partition_selector_fn instead of date_partition_range and create_default_partition_selector_fn for all the other schedule decorators.
Wed, Oct 14
Tue, Oct 13
Mon, Oct 12
Back to your queue
Thank you so much for doing this! This is a huge deal and will help all of our users
Sun, Oct 11
I feel not obvious/easy to get back to the playground/overview pages if you end up on a snapshot page, especially since we use the same tabs UI for both ["Definition, "Runs"] and ["Overview", "Definition", "Runs", "Partitions"].
Fri, Oct 9
This prevents accidental deletion of tabs and loss of configuration work.
Thu, Oct 8
I think this fixes the error @max ran into last week, but I'm not sure. Either way, doesn't hurt to do a mkdir - p
works locally but not in BK :/
I realized this is a user-facing error. This looks good, but I feel like we shouldn't be using check.**_param calls for things like this. The workspace_config argument is irrelevant to the user here. What do you think about doing an isinstance check (either in this function or before calling ensure_workspace_config) and throwing a DagsterInvariantViolationError or similar if we see that the config is not a dict.
They probably doesn't add that much of a performance hit, but I wonder if we should conditionally include the logTimeLink and timeStartLink themselves based on the feature flag.
We could add a debug console somewhere that allows toggling this kind of thing, then ask folks to toggle it on while we're live debugging with them.
Wed, Oct 7
Is this an error that we expect a user to see?
or is it something @max 's automated screenshot script can help?
Interesting, so I would be re-executing the *config* with the *current* pipeline, not the config with the *historical* pipeline. Is that correct? Should I be able to execute the historical pipeline instead? (Is that allowed?)
In the case of a historical config, would that mean that you get a config that might not work for the current pipeline at all but that that wouldn't be clear until you tried to run it?
should "Open in Playground" and "Re-Execute" be available in the context menu for old runs whose snapshots are now historical?
Unlink the pipeline name. This seems to be a point of confusion, where people want to click to the run itself, but instead click the pipeline name and end up on the pipeline instead.