Run conformance tests nightly on main branch

[sjberman]

In order to ensure we are staying ahead of Gateway API changes, we'll now run conformance tests off of the main branch on a nightly schedule.

Devs can also run from the main branch locally by setting the GW_API_VERSION variable.

Closes #668

• I have read the CONTRIBUTING doc
• I have added tests that prove my fix is effective or that my feature works
• I have checked that all unit tests pass after adding my changes
• I have updated necessary documentation
• I have rebased my branch onto main
• I will ensure my PR is targeting the main branch and pulling from my branch from my own fork

[pleshakov]

Hi @sjberman

is it possible to run the Gateway API conformance code from the main branch too? in addition to installing the main branch resources.

Looks like we're not on the same page about the requirements. My interpretation of them is to to run both the updated tests and the resources.

[sjberman]

@pleshakov Good call, I didn't even think of that.

[sjberman]

@pleshakov Ready for re-review

[pleshakov]

@sjberman
this looks good!

In the previous message #668, I missed one more requirements from that issue " forks from Gateway API." my apologies :(

basically, this is needed so that it is possible to run the tests from someone PR into the Gateway API repo. Like this one kubernetes-sigs/gateway-api#2179
this way, we can provide feedback (if any) and most importantly, hopefully can catch any non-compliant changes before they are merged to the main branch.

I was able to run tests using that fork by doing:

go mod edit -replace=sigs.k8s.io/gateway-api@v0.7.1=github.com/levikobi/gateway-api@attached-routes-conformance
go mod download
go mod verify
go mod tidy

instead of

	go get -u sigs.k8s.io/gateway-api@main
	go mod tidy

result:

        --- PASS: TestConformance/GatewayWithAttachedRoutes/Gateway_listener_should_not_have_the_route_attached_if_the_route_is_not_allowed_by_the_listener's_protocol (0.02s)
...

if that is a good approach, is it possible to enable it in the Makefile and documentation?

[sjberman]

@pleshakov I wonder if I document the steps you took, but don't create amake target for it? Since an argument is required (the path to your fork and branch), doesn't really make sense as a target for make. Could be a simple bash script, which I can write, but it's so simple I wonder if simply documenting how to do this is enough for developers, since I doubt this is a workflow that we'd be performing too often. What do you think?

[pleshakov]

@sjberman

Since an argument is required (the path to your fork and branch), doesn't really make sense as a target for make.

what if the default for replacement is github.com/kubernetes-sigs/gateway-api@main ? otherwise, the developer can specify the fork and branch they need

at the same time

@pleshakov I wonder if I document the steps you took, but don't create amake target for it? Since an argument is required (the path to your fork and branch), doesn't really make sense as a target for make. Could be a simple bash script, which I can write, but it's so simple I wonder if simply documenting how to do this is enough for developers, since I doubt this is a workflow that we'd be performing too often. What do you think?

seems reasonable too 👍

[pleshakov]

👍

[kate-osborn]

LGTM