I tweaked some code and ran a test a couple of revisions prior to the tip, accepting the changes. I then did an update, but bailed on resolving the test conflicts because there were too many, and simply reran the test. I totally forgot about the unmerged state until I tried updating elsewhere. But in the meantime, it let me (evolve) amend, and `hg ci --amend` (with evolve disabled). The working directory is clean (and summary agrees, though it notes one unresolved file, as does `hg resolve --list`). Vanilla commit and update are both blocked. I can see a selective amend (-I/-X/path) after an update being useful even if there are unresolved files, as long as the selected ones aren't unresolved. But allowing an unresolved file to be amended feels like a bug. If not, it should probably clear the unresolved state.
> allowing an unresolved file to be amended feels like a bug Yeah, sounds like a bug. I guess the amend would abort before e8a7c1a0565a "cmdutil: remove the redundant commit during amend."
Hey, I am not able to reproduce this. Can I get some clear understanding of how I can reproduce this?
okay, now I am able to reproduce this.
Fixed by https://mercurial-scm.org/repo/hg/rev/eeb87b24aea7 Yuya Nishihara <yuya@tcha.org> amend: abort if unresolved merge conflicts found (issue5805) It was checked by repo.commit() before e8a7c1a0565a "cmdutil: remove the redundant commit during amend." (please test the fix)
Bug was set to TESTING for 7 days, resolving