CI/CD combines continuous integration (CI) with either continuous delivery or continuous deployment, so the abbreviation can be ambiguous. Continuous delivery is the broader discipline; CI is one of its practices. A deployment pipeline is the pattern that moves software from commit to production and is often called a CI/CD pipeline. section 1.2.2
Continuous integration keeps integration work small and frequent. Developers make incremental changes, push them to the remote mainline at least daily, and receive early feedback on build health, tests, conflicts, and unfinished work. Incomplete features can remain hidden from users with feature flags or keystone interfaces while still being integrated. section 1.2.1
The workflow connects an inner loop to an outer loop. The inner loop covers writing, testing, running, and debugging a change locally. A push to the mainline closes that loop and triggers the outer loop, which continues until the change is deployed to production and operational. section 1.2.1 section 1.2.2
A deployment pipeline organizes the outer loop into three stages:
- Commit: triggered by every mainline push; compile code, run mostly unit and component tests, perform static analysis, and create a build artifact. section 1.2.2
- Acceptance: a named stage after commit on the path toward production. section 1.2.2
- Production: the final named stage in the path from a committed change to deployed, operational software. section 1.2.2
Fast feedback after a mainline push
A developer completes an increment, integrates it with the mainline, and lets the commit stage validate it. If tests fail, the developer immediately commits a fix or reverts the change. Keeping this stage ideally below five minutes avoids waiting, limits cycle time, and reduces friction. section 1.2.2 section 1.3.7
CI/CD depends on making mainline integration routine rather than a delayed event. Required pre-integration reviews can block work when reviewers are unavailable, while long-lived feature branches can extend feedback loops for days or weeks. Frequent integration reduces that delay and exposes problems sooner. section 1.3.7




































