I noticed a change of behavior when upgrading from 2.45.2 to 2.46.0. You can decide whether this is a regression or an improvement. I thought I’d say something just in case. Don’t shoot the messenger! Overview: The --summary option to `git diff --cached` seems to override the --exit-code option in 2.46.0 such that this is no longer true: > --exit-code > Make the program exit with codes similar to diff(1). That is, it exits with 1 if there were differences and 0 means no differences. You can reproduce yourself by staging something, then trying the 4 combinations of: - With and without --summary - 2.45.2 and 2.46.0 You will see that with 2.46.0 + --summary flag => exit code is always 0. This is different from 2.45.3 where --exit-code influences sets the exit code regardless of --summary More context: I have a scheduled script that daily updates submodules. The logic is: ```bash git submodule update --init --remote --checkout git add path/to/submodule if git diff --cached --exit-code --summary; then # Nothing to commit else # Create commit and push fi ``` This has been working for a while, most recently with git 2.45.2. After updating to 2.46.0, it stopped working. It would _always_ think there’s "nothing to commit", even when the submodule did have new commits. After investigating, I discovered that the --summary option seems to override the --exit-code option. i.e. if --summary is present, then the exit code is always 0, and not 1 if there are staged changed (as I’d expect from --exit-code manual entry) This is not limited to submodules, the behavior is the same for regular staged files, e.g.: ``` # Stage a file: $ git add debug.md # Diff with 2.45.2 prints summary and exits with 1: $ git diff --cached --exit-code --summary ; echo $? create mode 100644 debug.md 1 # Diff with 2.46.0 prints summary and exits with 0: $ git-2.46.0 diff --cached --exit-code --summary ; echo $? create mode 100644 debug.md 0 ``` I fixed my script by dropping the --summary option. But maybe this behavior change was unintended and not covered by tests, in which case you may want to treat this as a bug report. Taking this chance to thank you for reading and for maintaining git. I don’t know what I’d do without it! Cheers, M’
