Relationship with cache: are warnings returned in subsequent requests?

[ioggstream]

I expect

The spec to mention whether and how the "warning" attribute is part of the representation.
Is that supposed to be returned even by subsequent GET requests?
How does this impact with caching?

[andrecedik]

I'd say it depends on each individual case. I would think that the reasons for returning embedded-warning can be manifold. A system could e.g. return warnings due to circumstances in the backend that can either still exist on subsequent requests or that have been solved in the meantime.

So I'd say that caching of warnings shouldn't be done.

Do you have an example in mind where caching would make sense @ioggstream?

[andrecedik]

Hey @ioggstream if you have a look at 245804b and give me feedback if it's easy to understand I'd appreciate it.

Also, I'm still on the hunt for use cases that disprove the idea that caching of the embedded-warning content-warning type shouldn't be done. 😬
If we don't find any then it's safe to say, that we're on the right track 😉