Expansion of "how" section for Payload Processing #21
Conversation
|
|
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: shachartal The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
Welcome @shachartal! |
k8s-ci-robot
added
size/L
| # Indicates if this processor rule can mutate passed data | ||
| mutating: true | ||
| # Determines which parts of the request/response pair is of interest to the processor | ||
| phases: ["request-headers", "request-body", "response-headers", "response-body"] |
There was a problem hiding this comment.
This is reminding me of Istios EnvoyFilter, and Envoy proxy config for filters.
2 q's:
- Should we allow for trailers?
- Should we allow for streaming/buffering/chunked
There was a problem hiding this comment.
- What do you mean by trailers in this context?
- I think we need to allow processors to not materialize the full object before operating on it. We should take this into account as part of the Gateway-Processor protocol.
There was a problem hiding this comment.
gRPC is the most notable implementation to use HTTP/2 trailers https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Trailer
There was a problem hiding this comment.
Got it. Yes, I think we should extend this to account for trailers, since we do want to support gRPC payload processing.
| kind: Service | ||
| name: bad-prompt-detector | ||
| # Configuration for the callout itself | ||
| remoteConfig: |
There was a problem hiding this comment.
Is this field unstructured, in the golang sense?
These are arbitrary values passed as context to the endpointRef?
There was a problem hiding this comment.
That's an excellent question. I haven't gotten that far ahead in my thinking here. I suggest some parts are structured and generic, while some others (the context field perhaps) should be unstructured. Perhaps we should split structured and unstructured as different top-level fields. WDYT?
4 participants
Added some initial thoughts and a draft proposal for a CRD for Processors.