Re: [RFC PATCH 0/2] teach `worktree list` to mark locked worktrees

[Rafael Silva <rafaeloliveira.cs@gmail.com>]

On Fri, Oct 09, 2020 at 06:50:02PM -0400, Eric Sunshine wrote:
> >
> > Sorry for the bit late response.
> 
> Likewise.
> 
> >
> > Doing a little investigation on the code, it seems the machinery for checking
> > whether a worktree is prunable it seems is already there implemented
> > on the `should_prune_worktree()`.
> 
> Yes, when I mentioned that in [1], I envisioned
> should_prune_worktree() being moved from builtin/worktree.c to
> top-level worktree.c and possibly generalized a bit if necessary.
> 
> One thing to note is that should_prune_worktree() is somewhat
> expensive, so we'd probably want to make determination of "prunable
> reason" lazy, much like the lock reason is retrieved lazily rather
> than doing it when get_worktrees() is called. Thus, like the lock
> reason, the prunable reason would be accessed indirectly via a
> function, say worktree_prunable_reason(), rather than directly from
> 'struct worktree'.
> 
> [1]: https://lore.kernel.org/git/CAPig+cTTrv2C7JLu1dr4+N8xo+7YQ+deiwLDA835wBGD6fhS1g@mail.gmail.com/

Appreciate the tip, I will be working on the prunable annotations, verbose
and other information that was proposed previously for the "worktree list"
command.

> > Additionally, having the ability to see the annotation and the reason in
> > case you see the annotation seems like more complete work for the intention
> > of the patch.
> >
> > Unless you think that is better to start with the annotation, and some time
> > later addressing the other changes specified by [2].
> 
> Whatever you feel comfortable tackling is fine. The simple "locked"
> annotation is nicely standalone, so it could be resubmitted with the
> changes suggested by reviewers, and graduate without waiting for the
> more complex tasks which could be done as follow-up series. Or, expand
> the current series to tackle verbose mode and/or prunable status or
> both or any combination.
> 

Thanks. I've just resubmitted the "locked" annotation patch, as you said,
it's nice standalone and can be integrated and hopefully will be already
useful for other git users and soon (hopefully :) ) will submit new patches
for the other changes as proposed by [1].

[1]: https://lore.kernel.org/git/CAPig+cTTrv2C7JLu1dr4+N8xo+7YQ+deiwLDA835wBGD6fhS1g@mail.gmail.com/