Joy Reynolds
Joy Reynolds
> So taking the case of N=2, it makes sense that all of the properties would remain on the last commit in this case as well. So, this same logic...
The arrows make it look circular, which I doubt you intended. What would it look like for something like `gg` which simply wants to use jj-lib and not jj-cli?
From the 0.34.0 CHANGELOG: > Git-based repositories are now colocated by default. Configure `git.colocate = false` to keep the previous behavior.
What if the problem is that the `new -A` wasn't represented correctly in the original output? IMO, the `new` command creates a branch in the graph, no children. So the...
The test content shouldn't change just because a core theme doesn't do it correctly.
> could you describe in what situation you already have a `Change-Id` trailer. Which commands reprocess the description? (`rebase`, `split`, `new x y`, `squash` are some) Does a signing of...
> I guess if we're already relying on conventions like `trunk()`, then we could certainly also have a conventional revset that defines commits to ignore by default? It sounds dangerous...
Is this revset useful for anything outside of `log` context? _(I don't actually understand what it returns or why you would want it.)_ If it is only `log`, maybe it...
> t may be okay to stick with dirs as the more widely used one, and just override what we need. Wouldn't everyone need that? Submit the change upstream?
It seems that your key/value pair cache could easily be external (SQLlite?), as the pre-commit hook is also. You can't use the change ID for your example, since it can...