Why does a bitbucket pipeline on one branch try to merge with a commit from another branch

[Rutger Claes]

We have a build that failed with the following log during the "Build Setup" step:
```
+ git reset --hard 38167b31fd00
HEAD is now at 38167b3 DAT-2701: fix some thing

+ git config user.name bitbucket-pipelines
+ git config user.email commits-noreply@bitbucket.org
+ git config push.default current
+ git config http.${BITBUCKET_GIT_HTTP_ORIGIN}.proxy http://localhost:29418/
+ git remote set-url origin http://bitbucket.org/datylon/daty-vizlib
+ CONFLICT_EXIT_CODE=3
+ git merge 263e287ee12d --no-edit || exit $CONFLICT_EXIT_CODE
warning: Cannot merge binary files
```

38167b3 is the head of a feature branch.  263e287ee12d is a merge of that feature branch with our release branch.

So why does the pipeline setup for the feature branch merges in a commit that was made later?

[Deleted user]

We enountered the same problem. Our expectation was that the Pull Request build executes on the branch the pull request is from, and we don't want the build to use anything from the destination branch.

@Graham GatusThanks for the tip on using "branches" pattern, but this requires us to enforce using patterns when we create branches (e.g. /feature/...).

Seems Bit Bucket doesn't have any other option in this case and Pull Request pipelines becomes useless.

[Graham Gatus]

@Rutger Claes do you mind sharing a snippet from your bitbucket-pipelines.yml file?

It sounds like you have defined a pull-requests section in your bitbucket-pipelines.yml file.

If a branch matches the pattern defined for pull-requests, when a pull request is created, the destination branch will be merged with the source branch, and a pipeline run against the result. This can be used run tests against the potential result of a merge.

If you don't want this behaviour, instead of defining a pull-requests section, use a regular branches section instead. e.g:

pipelines:
  default:
  - step:
      script:
      - echo "This script runs on all branches that don't have any specific pipeline assigned in 'branches'."
  branches:
    feature/*:
    - step:
        script:
        - echo "This script runs only on commit to branches with names that match the feature/* pattern."
  pull-requests:
    feature/*:
    - step:
        script:
        - echo "This script runs only on commit to branches with names that match the feature/* pattern when a pull request is raised."

[Anton Daneyko]

This costed us about a day of investigation. The Web UI clearly shows which git hash it builds from. The fact that an additional merge happens is hidden within the "Build Step".

From the pipelines web page it looks like the build with the same git commit passes, and at some point it breaks. We suspected some of our tests were unstable (non-deterministic) and it took considerable amount of time to figure out that the code that was tested does not come from the hash that is displayed in upper left corner of the Web UI.

[Deleted user]

Why does it do this? If I wanted to update my branch with respect to the destination branch, I would do that myself.