Hello, We are implementing bi-directional testing in some of our projects, but it's not clear to us how the contract matching is done between consumer versions, provider versions, and different environments. For example, is the environment matching one-to-one (i.e., consumer 'dev' with provider 'dev'), or are they separate? Can a consumer in a 'dev' environment match with a provider in 'production', for instance? And how do we control that? Thank you.

Hi!

but it’s not clear to us how the contract matching is done between consumer versions, provider versions, and different environments.

By default, we match against the latest version, and any deployed or released version of the reciprocal of the integration. e.g. If you publish version

abc123

of application

A

(in an

A

->

B

relationship) Further, if the latest version of

B

is

xyz098

, and version

abab2323

is deployed to

staging

and version

baba6565

is deployed to

prod

, then we check if

abc123

is compatible with: 1. The most recent version of

B

(

xyz098

) 2. All deployed/released versions a.

abab2323

(staging) b.

baba6565

(prod) And vice versa. When you call

can-i-deploy

for version that doesn’t have a compatibily matrix entry, we perform a comparison on demand. Use the polling arguments to wait for a result (usually a few seconds).