Stephen Day

@thaJeztah This is a little surprising. Why would we wait at all to run the healthcheck? The "start period" should be the time after which we consider a failure to...

@alexpirine Ok, it really should just start running and only consider a fault as a failure until after the start period to minimize bring up. We should consider this a...

> Is redis:3.2-alpine typoed? I'd expect ‘redis alpine:3.2’ to create > refs ‘latest’ (with the redis:latest content) and ‘3.2’ (with the > alpine:3.2 content). Read the example again. The tags...

@runcom Would love to see a WIP PR. Let me know and we'll work towards something.

@cpuguy83 > Definitely -1 on a --oci flag. I think --format oci would be better. What other formats do you see working here in the future? Ideally, we'd be able...

@duglin I don't think we'd want to require a flag per version. Once you know what class of format you're accepting, disambiguating the version is done with internal mediatypes. If...

> I was thinking more on the docker save side of things - where its a question of how does the user want the image serialized. docker save --format ociv1...

@duglin You missed my point. This isn't as simple as a format change. There are options and behavior that change. For example, `--ref` isn't valid. Also, you've failed to recognize...

@duglin Exactly. And we avoid the weeds if we just use flags, rather than a micro-format.

@mikebrow > Not sure why --format type is a micro format and --type isn't. Don't they hold the same meaning? `--oci` would be a documented flag. We have documentation about...