feat: support etag header for client-side eval

[markphelps]

Re: flipt-io/flipt-client-sdks#294

Support etag/if-none-match headers for client-side eval endpoint

~ » curl -v  http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default

* Host localhost:8080 was resolved.
* IPv6: ::1
* IPv4: 127.0.0.1
*   Trying 127.0.0.1:8080...
* Connected to localhost (127.0.0.1) port 8080
> GET /internal/v1/evaluation/snapshot/namespace/default HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/8.6.0
> Accept: */*
> 
< HTTP/1.1 200 OK
< Content-Length: 1401
< Content-Type: application/json
< Etag: 7cfafac21953b341569a2a676205e0c0e5ff7e00
< Grpc-Metadata-Content-Type: application/grpc
< X-Content-Type-Options: nosniff

---

~ » curl -v  -H "If-None-Match: 7cfafac21953b341569a2a676205e0c0e5ff7e00" http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default

* Host localhost:8080 was resolved.
* IPv6: ::1
* IPv4: 127.0.0.1
*   Trying 127.0.0.1:8080...
* Connected to localhost (127.0.0.1) port 8080
> GET /internal/v1/evaluation/snapshot/namespace/default HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/8.6.0
> Accept: */*
> If-None-Match: 7cfafac21953b341569a2a676205e0c0e5ff7e00
> 
< HTTP/1.1 304 Not Modified

[erka]

@markphelps your code looks great but there is a problem. If you have the namespace with some amount of flags and segments http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default will produce different etag. I believe the logic uses map somewhere and it produces the different order of elements in the json.

Here is the sample

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 4bbaa020a72c5bd21b224bcdd11c5bf4c7c0cda3

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 92e8e67b990930eca494f3feb4536c68725dc2bf

[markphelps]

@markphelps your code looks great but there is a problem. If you have the namespace with some amount of flags and segments http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default will produce different etag. I believe the logic uses map somewhere and it produce the different order of elements in the json.

Here is the sample

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 4bbaa020a72c5bd21b224bcdd11c5bf4c7c0cda3

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 92e8e67b990930eca494f3feb4536c68725dc2bf

gah you are right. good find @erka . back to the drawing board it seems

[markphelps]

@markphelps your code looks great but there is a problem. If you have the namespace with some amount of flags and segments http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default will produce different etag. I believe the logic uses map somewhere and it produce the different order of elements in the json.
Here is the sample

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 4bbaa020a72c5bd21b224bcdd11c5bf4c7c0cda3

> curl -v http://localhost:8080/internal/v1/evaluation/snapshot/namespace/default 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Length: 7011
< Content-Type: application/json
< Etag: 92e8e67b990930eca494f3feb4536c68725dc2bf

gah you are right. good find @erka . back to the drawing board it seems

Its likely related to us using protojson to marshal the output which isnt deterministic: https://protobuf.dev/reference/go/faq/#unstable-json

we need to think of otherways we can produce a stable etag value

[markphelps]

closing for another approach