Hi all, i am wondering about how to properly use can-i-deploy and would like to hear your take on this_._ This question is for the Consumer side. Context: Contract is between • 2 mobile consumer • 1 web consumer • 1 provider • the mobile apps need to be back wards compatible for N amount of app releases • The Broker is set-up per feature, this means we will have multiple brokers in our cluster Pact tools to use • Can-I-Deploy • Publish • Recordings Currently we are implementing everything in the pipeline. The plan is to make sure all contracts are verified when we want to do a merge. We think we need to use Can-I-Deploy for this, but it feels like that is not how it should work. The reason we think that is because Can-I-Deploy seems to only work for 1 single pacticipant at the time, we know you can give multiple pacticipants like this.

pact-broker \
can-i-deploy \
--pacticipant=IOSConsumer \
--latest
--pacticipant=AndroidConsumer \
--latest
--broker-base-url=<https://pact-broker.com/feature> \
--to-environment=PROD

But it feels like a lot of maintenance to always have all your pacticipants up to date. So in short, which answer applies most 1. We are mis-using Can-I-Deploy. 2. We are doing it right and should take care of managing all the pacticipants on multiple brokers. 3. We Should run all pacticipants of one broker at once, and we missed some pact functionality

There are multiple things I think wrong here

Hi Matt, tnx for your responses and explanations. We are indeed moving away from --latests, so that was a poor example, we are now also looking into record release, since using tags is not advised. But i can't figure out a way to either, check all the pacts, always, which i think is not common. or "give it the thing we want to deploy" how to check if we can deploy in a CI/CD way without any manual intervention. Just to be sure and clear, we are talking about native mobile apps, not mobile web apps. This is why we want to check everything for everything and the release of all feature will have the same upcoming version. That is why i was saying it feels like we are not using can-i-deploy correctly. because now we require a script to ensure we have a list of all our pacts in all our broker so we can use can-i-deploy for all of them, instead of saying check all for this broker. But yeah, new to pact so perhaps we just have it completely wrong 😅

For each mobile app you wish to release, you should

pact-broker record-release

. Each one of these are now considered “in production”. When any dependency of that changes (e.g. the

provider

) and wants to release, when it calls

can-i-deploy

it will check all released versions of the mobile app. i.e.

can-i-deploy

figures out what’s in the target environment, and if the app being deployed into it will be compatible

okay, but then in my mind, my challenge shifts from testing all pacts, to finding out which pacts changed and put those through can-i-deploy. because one app will have multiple pacts and it's pacticipants. is that correct 🤔 Or is this just a manual task and should not be integrated into CI

okay, but then in my mind, my challenge shifts from testing all pacts, to finding out which pacts changed and put those through can-i-deploy. because one app will have multiple pacts and it’s pacticipants. is that correct

I don’t understand why you need to work out which pacts changed? If you’re deploying Application A, simply ask

can-i-deploy

if you can deploy it. Whether or not the contracts have changed is orthogonal to that problem. It will say “OK, you want to deploy

A

into

prod

, let me go and look at all of the dependencies of that in prod and check that you won’t break them and/or they won’t break you” (plus all released versions, in the case of mobile app and the provider is the one asking the question)

Well, we are working consumer driven, so we are planning to run

can-i-deploy

from the consumer side (the mobile apps). When you say

Deploying application A

, this means multiple pacts associated to that application right? Therefor the assumption remains that you need to run

can-i-deploy

for all your separate pacts that you have. How else do you make certain all is well?

I mean, you can run

can-i-deploy

(or

can-i-release

) for you deploy/release either side. You should do both, really.

We should use the provider pacticipants when we do

Can-I-Deploy

, not the consumer pacticipants.

yes! or rather, in the consumer build pipeline you add the consumer app that you are deploying. In the provider build pipeline you add the provider app that you are deploying. in general - you specify the pacticipant you are deploying in can I deploy. Think of saying it in English to your team mate “That work on ApplicationX is done, can I deploy it to production?” That’s effectively what can-i-deploy is for

Cool, sorry it took me that long to figure out. I will look into how to record-release to all pacts in CI I might be back with some more questions 🤭

No worries!