Skip to content

Extend API with GET evaluation request #613

@iamafanasyev

Description

@iamafanasyev

What are your thoughts on extending the existing evaluation API with functionally equivalent (to POSTs) GET requests (e.g. GET /evaluation and GET /evaluation/batch)?

Several pros could be observed from general perspective (like idempotency of evaluate), but the main actual reason I'm asking is that my client feature flag request is time critical, and:

  1. POST /evaluation/batch ...
    ...
    Content-Type: application/json
    ...
    
    is always not a simple request from CORS perspective, so a preflight request is always spawned (and it is quite important for us to have an option to keep our services on different hosts for some organisational purposes);
  2. there is no option to preload the request (by inlining it in the resulting client's HTML);
  3. GET request is more straightforward from cache control perspective (e.g. when your application shares the same feature flags across different "pages").

(at the moment we address these drawbacks at the infrastructure level: edges/lambdas, reverse proxies, etc.)

Cons are obvious I guess — URL length limitations. Any other?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions