Also, could you describe the current code-review process for Tangled in more detail please?
My belief is that code review should happen locally, and the unit of work being reviewed stays independent from unit of work being "submitted". The reviewer should be able - locally or in the WebUI - to specify a change-/revision-set (preferably via jj revset language), or add files to review ad-hoc, or even mark specific lines for review!
And then, assign comments to such a review unit - where changes are all being captured as a first-class object, with all the niceties that may come together, e.g.:
- comments are captured in the jj oplog as if they were code changes
- it's easy to surface (locally or in the WebUI) past code reviews
- it's easy to "untangle" and grasp which particular comment belongs to which code review
- [vague from my side] comment might belong to multiple code reviews, or might not; code review might belong to multiple revsets
Also, could you describe the current code-review process for Tangled in more detail please?
My belief is that code review should happen locally, and the unit of work being reviewed stays independent from unit of work being "submitted". The reviewer should be able - locally or in the WebUI - to specify a change-/revision-set (preferably via jj revset language), or add files to review ad-hoc, or even mark specific lines for review!
And then, assign comments to such a review unit - where changes are all being captured as a first-class object, with all the niceties that may come together, e.g.:
- comments are captured in the jj oplog as if they were code changes
- it's easy to surface (locally or in the WebUI) past code reviews
- it's easy to "untangle" and grasp which particular comment belongs to which code review
- [vague from my side] comment might belong to multiple code reviews, or might not; code review might belong to multiple revsets