Welcome @davidwin93!

It looks like this is your first PR to kubernetes-sigs/gateway-api 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes-sigs/gateway-api has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

Hi @davidwin93. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Thanks @davidwin93, this will be a great improvement!

/ok-to-test

/cc

Note: For historical reasons, a user agent MAY change the request method from POST to GET for the subsequent request. If this behavior is undesired, the 307 (Temporary Redirect) or 308 (Permanent Redirect) status code can be used instead.

Is there any suggestion that we would want to include in docs for this behavior for 301 and 302 responses?

I'd like to see the user-facing docs updated as part of this PR -- thanks!!

@LiorLieberman in regards to SupportedFeature do I need to wait for this to get in first?
Do you happen to have any examples of any in flight work that is currently using this functionality, or would this be the first change to leverage supported features?

this may benefit from a SupportedFeature for gradual enablement, because if this gets in, if there is any implementation that does not support 307 and 308 they go out of conformance

Can we add this as Extended (optional)?

@LiorLieberman in regards to SupportedFeature do I need to wait for this to get in first? Do you happen to have any examples of any in flight work that is currently using this functionality, or would this be the first change to leverage supported features?

no, these are two different things. when I said supported features I really meant a FeatureName - like https://github.com/kubernetes-sigs/gateway-api/blob/main/conformance/tests/httproute-response-header-modifier.go#L40 for example.

And @candita, to your point, adding a featureName would mean it is initially extended, but as said during the meeting, we should communicate clearly we are only doing it for gradual enablement, with intentions to move it to core (and remove the featureName) once we communicated it enough

@LiorLieberman this should be good for another pass. I believe I've addressed the comments correctly.

do we have some implementation already supporting these return codes, just so we can do a run and guarantee that conformance is also ok?

Thanks!

Thanks everyone for the work and discussion on this. I completely understand why this has taken some time to make it in, and also why it didn't make the most recent release. These things happen, and the process around release management and validation criteria is important to keep stable.

That said, I do want to highlight a concern from the perspective of downstream adopters and integrators. When changes like this can't make it into a patch or minor release—even when they're relatively small, such as expanding an enum or adding a bit of CEL validation—it effectively means that these improvements can take close to a year to reach users, given the six-month release cadence.

That long gap can make it difficult for projects building on Gateway API to support their users or align on conformance, especially when a missing option or validation rule blocks certain use cases. Over time, this can lead to the same challenges that the Ingress API experienced, where implementations resorted to annotations or custom extensions to fill gaps that the upstream API couldn't evolve quickly enough to cover. You can see in the reference above where we have moved to do exactly that in kgateway due to this not merging and that given prior discussion here our understanding is that changes like this can only go in minor versions.

For those reasons, I'd like to suggest we re-evaluate how we classify and handle changes like this. It seems like there should be a safe path for small, backward-compatible improvements to land in patch releases, especially when they improve expressiveness or correctness without changing behavior or compliance.

Really appreciate everyone's continued work here; I just want to make sure Gateway API can keep evolving fast enough to stay broadly usable across the ecosystem.

This is part of the point of the changes discussed in #4164 -- the idea is that yes, it should be faster to get things into an experimental-channel release that people can experiment with.

Thanks @kflynn! A very timely discussion, from a skim I think moving to a model of monthly releases for the experimental channel makes a lot of sense. To make sure I understand correctly, in the context of this PR where there is an agreed upon change that should be made to the stable & experimental versions simultaneously, it just means that the changes would be available at most within a month of merging via the experimental channel? Then they will be in the next stable release whenever that happens?

That said, I do want to highlight a concern from the perspective of downstream adopters and integrators. When changes like this can't make it into a patch or minor release—even when they're relatively small, such as expanding an enum or adding a bit of CEL validation—it effectively means that these improvements can take close to a year to reach users, given the six-month release cadence.

You're completely right, this took too long to get in. Improving this is a big focus for us, a big thanks to @kflynn for helping move this forward.

There was some concern if this was too late to squeeze in before v1.4, but now that that release is out, I think it's past time to get this change in.

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: davidwin93, robscott

The full list of commands accepted by this bot can be found here.

The pull request process is described here

I agree that this is a fine candidate for the next release! @davidwin93, one major upcoming change is that you need to line up some implementations who're willing to implement this as an experimental feature, although we, uh, haven't decided exactly how to record that. A comment here will do for now. 🙂

Istio and kgateway implement this experimental feature now.