7.0.0: Shallow clone by default
Highlights
The step now tries to do a shallow clone of the repository by default (instead of fetching the entire history) to speed up clones.
This means that you no longer need to define the clone_depth
input, it's only used for overriding the default behavior. By default, the step will shallow-fetch (depth=1) in all cases except for PR builds where the PR merge reference is not available. In this case, it fetches the last 50 commits and merges the PR branch into the target branch manually. You can use the clone_depth
input to set a higher depth value if needed.
Breaking changes and migration
- The step no longer fetches the entire repo history: if you have custom scripts or steps that rely on the commit history (or the
git log
command), you might want to set a customclone_depth
value. - Removed
manual_merge
input: this is no longer needed because the step prefers fetching the merge ref or the patch file and only does a manual merge if the former is not available. - Fixed incorrect spelling of output names:
GIT_CLONE_COMMIT_COMMITTER_NAME
,GIT_CLONE_COMMIT_COMMITTER_EMAIL
- Removed
PR_ID
input, although it didn't have any effect on the behavior.
What's Changed
- Use v2 stepconf parser in #189
- Remove
manual_merge
input, only do it as a fallback in #188 - Refactor output exports in #190
- Replace usage of COMMITER with COMMITTER in #192
- Add method for determining the ideal default clone depth based if the… in #191
- Use PR head ref in output export in #193
- Disable git GC before fetch #194
- Reorder inputs and categories in #195
- Remove PR ID input in #196
- Better step input descriptions in #197
Full Changelog: 6.2.3...7.0.0