Temporal coupling

Temporal coupling is a form of coupling in which correct behavior depends on the relative timing or ordering of operations, rather than on any structural dependency between the parts involved. Two elements can be temporally coupled even with no shared data or explicit reference between them, simply because one must happen before, during, or after the other for the system to behave correctly.

At the level of a class or API, temporal coupling is an implicit relationship between two or more members that requires a caller to invoke one before the other. For example, a connect() method that must run before send() is safe to call. This as a design smell because the ordering requirement is invisible in the members' signatures. Nnothing in the type system stops a caller from getting the sequence wrong.

The same idea appears at the level of distributed systems, where temporal coupling means that communicating processes must both be available at the same time for an exchange to succeed. This is the defining cost of synchronous request-response communication. The caller cannot make progress until the callee replies, so a slow or unavailable callee stalls the caller too.

Asynchronous alternatives, and patterns such as a message queue sitting between the two sides, exist specifically to remove this dependency on both parties being up at once.