monotone autotags

[Christopher Stafford]

Tim Ringenbach wrote:
> This email is a feature request for monotone, that I'm running by you
> guys instead, to see if it's actually a feature we would be interested in.
>
> Monotone's SHA1 revision numbers seem less than useful to me. Sure,
> they're great at uniquely identifying a revision. But they're
> impossible to remember, which makes talking about them difficult. If
> someone asked me a question about revision
> c4af5d58ae0a812912ffe29dc285f3d08afb5425 I would have no clue what
> they were talking about without looking at my commit emails or running
> a mtn command. Furthermore, I would have to copy and paste because I
> wouldn't be able to remember that number long enough to type it in.
> (Actually that's not true, both email search and monotone will let you
> enter just the beginning of it).
>
> So I was thinking, what if we tagged reach revision with something
> actually memorable. I have no idea if that would hurt monotone's
> performance or anything like that. Thinking further, I wouldn't want
> all those tags showing up with the release tags, so monotone would
> need some way to flag them as less important. In addition, it would be
> cool if monotone would automatically add a prefix and postfix to the
> tag. Hence I'm calling this potential monotone feature request
> "autotags". I picture it working something like this:
>
> mtn ci --auto-tag whitespace
>
> That revision gets tagged with something like
> marv at pidgin.im_whitespace_3 (pretending I had used that same tag twice
> before).
> Perhaps it could be a line to fill out in the commit template in vim,
> instead of a command line argument.
>
> I think this scheme makes the tags mostly unique, but since they're
> just aliases for the sha1's, it's not the end of the world if they're
> not. (How does monotone handle duplicate tags, btw?)
>
> What does everyone think? I'm running it by you all first before I
> even think of actually filing it as a feature request to monotone.
> There's no point in me requesting such a feature if you guys don't
> even think it's sane/useful.
>
> --Tim

Makes sense to me, although I think I'd lose the committer_id off the
start, since you could just report that in whatever places you reported
the auto-tag, and people would just refer to tim's change 'whitespace_3'.

chris