see https://docs.pact.io/wip and https://docs.pact.io/pending

or the CI/CD workshop 👇 (howtolearn)

(this one does block a provider branch since a roll out with a red job here means breakages)

You don't actually have to block the build. You just have to block deploy.

this problem stems mostly from the fact that we can’t do webhooks (networking security stuff); otherwise we’d have a way of verifying with a branch name coming from the env for instance

in the guides/workshops I couldn’t find/see where exactly one can configure the pact selector to include all consumer versions (perhaps only check latest of every version), would you mind pointing me to the docs for this? Additionally, I was reading about the pending pacts features, but I’m unsure that’s what I want

wait, maybe it is exactly what I want

I want in my CI/CD to only fail if mainbranch/deployed contracts can’t be verified, everything else I’m happy to just publish the result

ok, it seems I’m not quite understanding how to go about this. I enabled

pending=true

like so

enablePendingPacts = "true", providerBranch = "master"

for pact-jvm, and then I ran my provider tests. I know there’s a consumer branch for consumer X with contracts that can’t be successfully verified (for various reasons), however it still didn’t pick up those contracts during the run. Re-reading the docs for

pending

it seems that all this does is for the main branch/deployed allow for failure in cases when the consumer pushed to their main branch/deployed contracts that can’t be verified. And the point is to not block the provider from rolling out. This is good stuff if I got it right, but it doesn’t address my issue of “publish verifications for all contracts out there”.

@Timothy Jones you said “You might want “verify all contracts under a certain age”, plus main + deployed (edited)“, and this sounds about right. How is this achieved? And with pact-jvm, do you know?

I just ran my provider tests with

includeWipPactsSince = "2023-01-01"

, and it seems like this did the trick

the result was, in my IDE: 1) contracts for the branch I wasn’t verifying before were included in the run 2) the result of the tests was still

PASS

in the IDE 3) I didn’t publish because IDE but I am hopeful that when this runs in CI/CD a

failed

result will be published

I think you might need wip pacts too? To be honest I’ve never fully understood how pending works. I believe the answers are in the CI/CD workshop linked off the pending page, but I didn’t want to do a workshop when I last needed to know.

I think pending just means that a contract that has never passed is allowed to fail.

WIP pacts are here

It looks like the documentation is better than when I last looked 🙌

I just re-read your last message. It sounds like you got it to work!

thanks for your patience 🙂

this problem stems mostly from the fact that we can’t do webhooks (networking security stuff); otherwise we’d have a way of verifying with a branch name coming from the env for instance

An alternative is for us to have a CI job that takes the branch+consumer name from an env var, and we ask consumers to trigger this job after their publish-contracts job. It’s a bit like webhooks except we don’t go via the broker.