Add API Gateway Local setup

[rustworthy]

Addresses #196

[jonhoo]

This is excellent, thank you!

[jonhoo]

It'd be nice if we could somehow avoid the duplication between this and

wewerewondering/infra/apigw.tf

Lines 58 to 88 in 77525cb

	resource "aws_apigatewayv2_route" "api_event_post" {
	api_id = aws_apigatewayv2_api.www.id
	route_key = "POST /api/event"
	target = "integrations/${aws_apigatewayv2_integration.www.id}"
	}
	resource "aws_apigatewayv2_route" "api_event_eid_post" {
	api_id = aws_apigatewayv2_api.www.id
	route_key = "POST /api/event/{eid}"
	target = "integrations/${aws_apigatewayv2_integration.www.id}"
	}
	resource "aws_apigatewayv2_route" "api_event_eid_get" {
	api_id = aws_apigatewayv2_api.www.id
	route_key = "GET /api/event/{eid}"
	target = "integrations/${aws_apigatewayv2_integration.www.id}"
	}
	resource "aws_apigatewayv2_route" "api_route" {
	for_each = {
	get_eeq = "GET /api/event/{eid}/questions",
	get_eeqs = "GET /api/event/{eid}/questions/{secret}",
	post_toggle = "POST /api/event/{eid}/questions/{secret}/{qid}/toggle/{property}",
	get_q = "GET /api/questions/{qids}",
	post_vote = "POST /api/vote/{qid}/{updown}",
	}
	api_id = aws_apigatewayv2_api.www.id
	route_key = each.value
	target = "integrations/${aws_apigatewayv2_integration.www.id}"
	}

, but I don't know if there's an easy way to do that? I suppose we could do the terraform configuration somehow with SAM as well?

[rustworthy]

You think we could open an issue for that so that if the number of endpoints grows and we feel burden, we can optimize this.

I am actually always using the axum app without lambda layer plus DynamoDB Local when testing stuff end-to-end locally. It just works as you point out here

I can see though how having API Gateway Local setup might be useful - we can check that the mapping really works and that the request reaches axum app as expected.

[rustworthy]

and also I really love how things a described and organized in infra with Terraform

[jonhoo]

Yep, opening a ticket to do that as a follow-up seems reasonable to me!

Honestly, the other path we could take is to just have a custom check in CI that the two match up.

[jonhoo]

The next question now is: can we get CI to also run (and succeed with) cargo test --release?

[rustworthy]

This is excellent, thank you!

it is my pleasure, as for this thread, since we are talking to the Backend directly in the test rather via a server, the test binary compiles and succeeds in release mode right now:

So we just need to add the flag on the CI, I will open a chaser.

But I think what you mean here, is more about how do we test the lambda-power app (i.e. the LambdaLayer'ed app in release mode). I think this implies launching lambda_http::run(app).await? in the tests, building cooresponding API Gateway events for each route, acting and making the assertions.

The ultimate test though would imply launching the wewerewondering client and the server in release mode behind API Gateway Local (and talking to DynamoDB Local), then acting on the client side and making some assertions (we already got this thread)

[jonhoo]

A bit of all of the above. I'm arguably mixing up a bunch of different thoughts here, so let me put them in a new issue. Give me a sec.

[jonhoo]

#202

[rustworthy]

#202

Sounds like a plan.

I remember I touched (or rather pulled?) fantoccini previously for a leptos-powered app testing.