Handling repeated text annotations

[Robert Collins]

On Thu, 2008-02-14 at 14:26 -0600, John Arbash Meinel wrote:

> |> right/left-wins might produce more conflicts than a heads() based
> solution.
> |
> | We don't currently perform true annotate merges.  --knit just uses
> | newness information: "Was line x present in any ancestor, or was it
> | introduced by this revision?"
> |
> | So if you want, I can think about it, but I don't think it's
> strictly
> | relevant.
> |
> | Aaron
> 
> Well, this would still effect that. Because if 2 sides introduce the
> same text,
> heads will say that it *was* introduced in THIS (because there was a
> disagreement between left and right about this line.)
> right/left-wins will say it was only from an ancestor.

I think heads() will give meaningfully different results when staircase?
merge patterns are present - that is when a line goes
A
B
A

in one branch
and in the other 
AB

I can imagine left-wins and right-wins cases where the second A in the
first branch is annotated with the original A's revision, but with heads
it would have a new introduction line.

So for merge this may make no odds today; but for annotation we'd be
completely skipping the fact that it toggled to B, and thats IMO a bug.

-Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20080215/b886f4cf/attachment.pgp