-
Notifications
You must be signed in to change notification settings - Fork 40
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
omdb could interpret status of blueprint tasks #6440
Conversation
Some example output for
Some example output for
Before the initial blueprint has been loaded:
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This generally looks good to me! This will need some updates as part of #6399, so I'll base that on top of this.
Err(error) => eprintln!( | ||
"warning: failed to interpret task details: {:?}: {:?}", | ||
error, details | ||
), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Wondering if there's a way we can incorporate a check that this deserializes correctly. I guess to do that we'd need a test which kicks off a blueprint execution and ensures that it deserializes correctly.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We could definitely do more to reify the interface between tasks' status messages and omdb. In an ideal world they'd show up in the OpenAPI spec and you wouldn't have to do this at all. When we added background tasks the obvious way to do that was with an enum with variants for each task, but that seemed unwieldy on a bunch of levels. But yeah, the more we add here, the more annoying this is.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh yeah I think the common layer should be serde_json::Value, not an enum. But the last and first things we do can be to work with this more specific type.
I had to do something similar for the update engine—in there, I implemented a scheme where each event report carries the name of the spec it is associated with. That worked reasonably well. (It's similar in spirit to storing a type ID in dynamic objects that you can downcast to.)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
(If this is too hard feel free to defer this, I'm going to be making some changes in this area soon)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now that you mention the idea of a separate name specifying the schema, I wonder if we could first-class that in the JsonSchema (i.e., "if this string property is X, then this other property is of type Y") so that on the other end Progenitor could deserialize to the specific type.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That's basically a tagged enum, right? Except we're modeling it as a type rather than an enum variant. (The difference is that a tagged enum is a closed universe and a "type ID" string is an open universe)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, and maybe the answer is to have the JsonSchema for this type expose it the same way it would a tagged enum?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think I understand what you're saying, but I guess I'm not quite following what additional support Progenitor could provide over doing this by hand. (But let's discuss this separately.)
I opted not to do #6421 here.