To rerwite history or not - that's the question

[Kevin Bullock]

On 21 Feb 2013, at 10:14 AM, Isaac Jurado wrote:

>  - In Git there are quite a few ways to rewrite history, as part of its
>    massive combinations of commands and options, which could also
>    induce you to think that history rewriting is allright.

Minor nit: Git mostly has just one command that hides a number of different ways of rewriting history: git-rebase. Pretty sure we actually have more -bundled extensions- in Hg that provide different rewriting methods than Git has built in.

> However, many advertised workflows for Git actually encourage the
> opposite: extensive branching.  Some times even forking linear history
> on purpose [1].  Also, the fact that Git has octopus merges, even though
> almost nobody uses them, could also be interpreted as an aid to
> multi-branching.
> 
> [1] http://nvie.com/posts/a-successful-git-branching-model/

Good point, I'd forgotten about this. IIRC, the model referenced actually recommends merging with ancestors just to avoid "fast-forward" merges -- so that you can always see when a sync between repos happened. Hg prevents you from merging with an ancestor in most cases.

pacem in terris / мир / शान्ति / ‎‫سَلاَم‬ / 平和
Kevin R. Bullock