Python Goes Mercurial

[Lawrence D'Oliveiro]

In message <mailman.3064.1238560625.11746.python-list at python.org>, Terry Reedy wrote:

> Lawrence D'Oliveiro wrote:
>> <http://arstechnica.com/open-source/news/2009/03/python-adopts-the-mercurial-version-control-system.ars>
>> 
>> So what were these "strong antipathies" towards Git, exactly?
> 
> The relevant PEP is http://www.python.org/dev/peps/pep-0374/

Useful link, though it seems incomplete.

Interesting phrase "To make up for svn's lack of cheap branching"--it's not the branching that's the problem in Subversion, it's the merging. :)

> Git apparently allows 'changing history' by changing the underlying
> directed acyclic graph.

Every object in a Git repository is identified by the SHA-1 hash of its entire contents. For a commit object, that includes the references to its parent commit(s). Thus, 
there's no way <http://wlug.org.nz/VeryImpossible> to change these while preserving the identity of the commit object.

Yes, it's easy with the low-level Git commands to go back and synthesize past history, but you can't hide the fact that you've done so. One of the design goals of Git 
was guaranteeing the integrity of the entire commit history.