A repository may have multiple heads, which is useful in at least two cases:
As part of a merge (generally temporary)
- Bugfix versions (generally permanent)
At any time, you can view the head(s) of a repository with the command hg heads.
Not all SCM tools support ongoing multiple heads, since it is possible to achieve the same results with multiple repositories instead. However, especially if you are used to CVS, multiple heads may be more familiar or convenient.
First, the changesets to be merged are pulled from the source repository into the destination repository. At that point, the original tip is still a head, but the tip of the source repository is also a head. Normally this case of multiple heads is only temporary, as the two heads are usually joined back into one right away, by performing a merge.
See also: MergeMultipleHeads
Let's compare a scenario with and without multiple heads. Imagine a project releases version 1.0, and then 2.0, and then discovers a critical bug in 1.0 that must be fixed. Current work on the upcoming 3.0 version is ongoing.
Without multiple heads, you would clone the repository each time you tag a release. To work on version 1.1, you would clone the 1.0 repository and work on that, separately from the repository you are using for current development. One drawback of this approach is that you can't directly access the 1.1 work while you are working on the current code.
With multiple heads, you can perform 1.1 work inside the current (pre-3.0) repository. Just use Update to populate your working directory with the contents of the 1.1 tag, make your changes, and commit. To continue working on pre-3.0, just use update again to reset your working directory to that code base. If you want to pull those changes into other branches, one way to do it is with the transplant extension.
Multiple heads allows a single repository to contain the entire history of a project, including bugfix branches.
If a branch descends from a tagged revision, the whole branch inherits the tag. Briefly, if you tag a commit "A", each descendant of A will inherit "A" as a "branch name" until it gets a different tag. Descendents of a merge receive the union of names from each side.
So you can do:
$ hg heads -b changeset: 1704:c2755eba8631 tag: tip branch: 0.8 user: Johannes Stezenbach <js at linuxtv.org> date: Mon Feb 6 17:35:22 2006 -0600 summary: display revision numbers
..and see that you've got one branch in your tree, descended from something tagged 0.8. (Try 'hg heads .' in hg 2)
And you can check out that head with:
$ hg co -b 0.8 Using head c2755eba8631 for branch 0.8
See also: NamedBranches