Experimental implementation of liquid-hg

[Kevin Bullock]

On 18 Jan 2011, at 1:48 PM, Martin Geisler wrote:

> Matt Mackall <mpm at selenic.com> writes:
> 
> Hi everybody,
> 
>> The whole point of liquid is for people to avoid accidentally
>> modifying or deleting or rebasing publicly-visible changesets.
> 
> I also discussed this with Pierre-Yves and the others at the Zurich
> mini-sprint, but let me bring it up here again: do you think rebasing
> publically-visible changesets is a big problem?

Yes. A rebase is not a simple, fully-automatable (in the general case) operation. The changesets resulting from the rebase therefore might contain entirely different changes than the original changesets. Conceptually, you're still deleting existing published changesets when you rebase, and publishing new, different ones.

For this reason alone --force should be required to rebase frozen changesets, but having extra heads show up that you thought you'd deleted would be nice to avoid as well.

> It's an honest question, I don't feel that we get many questions here or
> on IRC of the form "Oh no, I just rebased something after pushing it to
> my server! What should I do?"

Hopefully this is simply because (unlike some VCSs *ahem*) hg has done a good job of discouraging changing history.

pacem in terris / mir / shanti / salaam / heiwa
Kevin R. Bullock