Updated tags support

[Robey Pointer]

On 28 Dec 2005, at 8:44, John A Meinel wrote:

> Johan Rydberg wrote:
>>
>> Another issue; should a tag always be available in the revision it
>> points out?  In other words; should tag FOO, that is a symbolic name
>> for revision BAR, be present after a "bzr branch -r FOO foo"?
>>
>>> From my point of view, the obvious answer is "yes".
>
> That is my specific problem with Niemeyer's implementation. Tags  
> are now
> subservient to revision ids, even though they are meta-information
> *about* revision ids.
>
> I would rather see tags independently revisioned. So set-tag,
> remove-tag, etc would create a new unique identifier in tags.weave,  
> with
> its own ancestry separate from the branch's ancestry.
>
> Here is a mockup:
>
> .bzr/tags # Current tags
> .bzr/tags.weave # Ancestry of tags
>
> bzr set-tag # Updates .bzr/tags
> bzr show # Prints out the list in .bzr/tags
> bzr show -r something # Prints out the history of tags.weave
>
> Now, the only problem with this setup is we now have a 'tags' file  
> which
> is the working set of tags, which needs to be saved to the weave.
> We could use something like:
> bzr commit-tags
> To just commit the tag changes. Optionally, 'bzr commit' could also
> check to see if the tag file needs to be committed.

I've been trying to stay out of this one, but I can't resist any more.

This all seems ridiculously complex.  Commit a tag? What? :)  Could  
we simplify this down?  If I have to type 3-4 commands to do that  
operation, I'll probably avoid the feature because I won't trust  
myself to remember exactly how to do it right.  Not everyone using  
bzr will have spent a year working on a source-control project.

Proposal for simplifying tags:

* Each tag refers to exactly one revision.
* When pull/merging from another branch, the other branch's tags are  
merged in.  For identical tag names, the other branch "wins".   
Optionally remove all tags that refer to revisions you don't have.

That's it.  Is there some use case that needs more complex behavior?

robey