I'd like to see it in action - I don't have a great sense of how this actually gets used and adding a test to describe that behavior would be great.

Do we have a hypothesis for why this would cause issues this week but not last?

In D9021#234129, @sandyryza wrote:

@jordansanders how strongly do you feel about stars vs. dashes? I don't mind changing this time, but fairly annoying if that needs to be another step in the release guide. (the stars are generated by the Quip markdown export).

Let's definitely remove the "Added information" line 😄

Do we support termination? Dagit makes that api call if so

rebase

Return false if can_terminate can't look up any tasks

Remove unnecessary dagit permissions (these were only included because https://dagster.phacility.com/D8978 hasn't landed yet)

Add StopTask permission to dagit and add a sleep to the solid so it's easier to manually test

In D8910#231664, @johann wrote:

users might still run into this issue when their image contains an ENTRYPOINT.

in docs?

This is a really good change - we should more explicitly test it and be a little bit more defensive w/r/t passing strings around.

rebase

rebase

Fixup diff description

Take it with a grain of salt, but it looks like we backed away from pre-commit hooks once before as well:

I don't personally love git-hooks and find them to often be more of a hinderance than a help: https://www.thoughtworks.com/en-us/insights/blog/pre-commit-don-t-git-hooked

rebase

update

update

update

s/metadata/stub_ecs_metadata

Use the new instance_cm fixture instead of the build_instance fixture (which no longer exists)

Rebase

Rebase

Rebase

Refactor to return a contextmanager

Rearrange stack so this comes first

Add comments

rebase

rebase

In D8573#224314, @dgibson wrote:

test failure looks unrelated?

In D8490#222107, @sandyryza wrote:

Did a user come across this? I didn't think we exposed OutputManager in our docs at all. I don't think we need it around anymore.