Re: [Monotone-devel] Re: Branch naming conventions

[Oren Ben-Kiki]

On Sunday 12 December 2004 09:14, Brian P. Campbell wrote:
> I realize it's a little ugly, but one branch naming convention that
> would not have to reverse domain names and would be in a natural (big
> endian) order would be to use URIs to identify branches.

But:

On Sunday 12 December 2004 10:00, Nathan Myers wrote:
> Incidentally, it was a big mistake in Java to use domain names at
> all, just as it's an equally big mistake to use any trademarked name
> in a C++ namespace name.  Software changes ownership frequently,
> domain names moreso.

There's a URI scheme that handles this problem (http://www.taguri.org/). 
You could use that if you are worried about domain names changing 
hands. Of course, like Brian said, it is rather ugly.

The branch naming problem is similar to the revision id problem. You 
have an ugly but unique id for guaranteed identification (a hash for 
revisions) but also need shorter friendly names (a tag for revisions).

So, why not use the same mechanism for both? Make branch == the root 
revision. The globally unique id would be the revision hash. For a 
friendly branch name, allow any tag of the root revision to be used as 
an argument for the --branch option. Maybe even go for broke and allow 
any tag of _any_ revision in the branch to be used as the argument. 

Either way, people get to pick and choose whatever suits their fancy as 
the friendly branch names. Any problem this mechanism might encounter 
(collisions etc.) will also occur in the context of revision tags; 
solve it once for both uses.

Have fun,

 Oren Ben-Kiki