[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Mercurial vs. git

From: Sergiu Ivanov
Subject: Re: Mercurial vs. git
Date: Wed, 23 Sep 2009 14:19:05 +0300
User-agent: Mutt/1.5.20 (2009-06-14)


On Wed, Sep 23, 2009 at 09:19:09AM +0200, Arne Babenhauserheide wrote:
> Am Dienstag, 22. September 2009 00:37:57 schrieb olafBuddenhagen@gmx.net:
> > I really think this is the most important feature of git. No matter
> > how much you screw up, you can always go back
> Though in practical work the ability to revert is similar in
> Mercurial. You might not have all changesets lying around, but
> there's always another source to get the original version from and
> add the changes you really want. That's a reason why many Mercurial
> users go for feature clones: If you want to implement a new feature,
> you just clone your repo, do your changes, and push them back when
> you'r satisfied with them.
> Since the cloning uses hardlinks for the history, the additional
> clone doesn't cost much time.
> As far as I know, people who use git rather go for seperate in-repo branches 
> (which already wrecked my own repo twice, because I did *something* wrong). 

Yep, this statement is right at least in my case :-)

However, I'm not sure that having clones has much advantage over using
private branches in this situation, because if you do break something
in *your* domain (either private branch or explicit Mercurial clone),
you can always go back both in git and Mercurial (at least such is my
understanding).  Problems might occur when you try to merge your
changes or do something that concerns the whole repository, the former
being much more frequent than the latter, IMHO.  I cannot see any
advantages of one way over the other when merging.  OTOH, when you
change something repository-wise, I'm not sure that you are much
better in Mercurial, because if there is only one instance of history
for all feature clones, you can always screw things up globally. (I
hope I understand things correctly.)
> > that you can always recover, you can go ahead an try all kinds of crazy
> > or otherwise "risky" stuff, which you'd never attempt otherwise.
> My approach with git is "I can get a new clone" ;) 

I'd guess this is inspired by the general Mercurial feature-clone
approach :-) I've only used git clone to get the repository for the
first time and then played with private branches to check things this
summer :-)
> That's why I much prefer git to subversion: regardles of what I did
> wrong, if everything breaks, I can just copy my changed files into a
> new clone...

I haven't used subversion (which is probably unfourtunate) but I think
I would have quickly gone crazy if I couldn't copy changed files into
another clone.
> > Only once you understand this (and of course also know enough about the
> > available possibilities to make use of it), you can truely appreciate
> > the power of git. Unfortunately, few people bother to go that deep :-(
> It's sadly not easy to get there. 

Is it much easier to become a Mercurial guru? :-)

reply via email to

[Prev in Thread] Current Thread [Next in Thread]