Recently I switched from SVN to Mercurial. Now I wonder how to realize my intended branching work flow in Mercurial according to good practice, hoping other developers understand what happens in the repository.
This is the work flow:
- Usually I have a trunk/default branch where work on the current release series happens. Let's say that is 1.x. At the same time I use a branch 2.x to work on the next major release. Changes in this branch may be radical, so merging with the trunk/default/1.x branch makes no sense here.
- After a while work on 2.x may be finished and version 2.0 gets released. Now I want the 2.x branch to be the new default/trunk branch and the current default/trunk to be the 1.x branch.
- Repeating this process, there may come a new 3.x branch. As before, if 3.0 gets released, 3.x should become the new default branch while the then current default should become the 2.x branch (again).
My question is not whether this work flow is a good one (I guess it's not fundamentally wrong). My questions is whether the way I realize this in Mercurial can be seen as good practice or if there are better opportunities.
So here is how I plan to manage branches in Mercurial ...
Starting from a repository with a single branch which holds the code of the current release series 1.x:
$ hg init
$ echo "hello world" > file1.txt
$ hg ci -A -m "Initial commit of 1.x code"
Start working on release 2.x:
$ hg branch 2.x
$ hg ci -m "Create new branch for 2.x development"
$ echo "Big new feature for 2.x" > file2.txt
$ hg ci -A -m "Add big new feature"
Meanwhile, do some work in current release series (1.x):
$ hg up default
$ echo "Minor adjustments specific for 1.x" > file3.txt
$ hg ci -A -m "Minor adjustments"
After some time release 2.0 is ready, yippee! Make default branch to 1.x and 2.x to default:
$ hg up default
$ hg branch 1.x
$ hg ci -m "Make default branch to 1.x branch"
$ hg up 2.x
$ hg ci --close-branch -m "Close branch 2.x"
$ hg branch --force default
$ hg ci -m "Make former 2.x branch to new default"
Now create a new branch 3.x and work in it, also work on default. Again, after some time 3.0 is ready and it's time again to manage branch names:
$ hg up default
$ hg branch --force 2.x # (reuse previously closed 2.x branch name)
$ hg ci -m "Make default branch to 2.x branch"
$ hg up 3.x
$ hg ci --close-branch -m "Close branch 3.x"
$ hg branch --force default
$ hg ci -m "Make former 3.x branch to new default"
The repo now may look like this ('o' are the heads):
o Branch default (3.x)
|
| o Branch 2.x
|
| o Branch 1.x
|
|
.
The main point I'm not sure about is if reusing branch names and juggling with the branch name default is good practice.
A lot of text for that question - sorry - but I wanted to be clear about what I'm doing.
See Question&Answers more detail:
os 
