Hello Björn,
On Mon, 3 Aug 2009, Bjoern Michaelsen wrote:
About the documentation: I really dont think you are better off than git
documentation-wise.
I was not comparing docs Mercurial vs. Git. I just expressed that
0) strip is lacking features
1) mentioned features (bookmarks, strip etc.) do not belong to core yet,
2) there is no description in the book (most likely because of 1))
By my definition, a feature is mature if it does belong to core, it is
well documented in the core docs, it provides the required functionality
and it is bug free. YMMV.
However, I have no doubt that the situation can be improved.
Regarding docs:
I appreciate the Mercurial book and have even provided some fixes for it
earlier, but as far as I remember it lacks descriptions of the most
popular extensions except MQ.
On the other hand Git has now some real paper book, e.g. "Pragmatic
Version Control Using Git", and some real good talks like Scott Chacon's
<http://www.gitcasts.com/posts/railsconf-git-talk>, general screencasts at
<http://gitcasts.com>, also there is the 'Git community book'
<http://book.git-scm.com>, 'Git in a Nutshell'
<http://www.chem.helsinki.fi/~jonas/git_guides/HTML/git_guide/> and 'Git
Magic' <http://www-cs-students.stanford.edu/~blynn/gitmagic/index.html>
(all work in progress).
I don't know if similar example screencasts are available for Mercurial.
It is just that with git you are required to understand even the more
advanced topics for basic operation. To do these non-basic operations,
you need to read and understand a bit more than a basic mercurial
tutorial (but not more than you did for git). For a
"hey-I-just-wanna-do-what-I-did-with-CVS"-dev, mercurial docs are much
easier.
As already said, I prefer Mercurials UI over Git's with the exception of
this local branch stuff.
In the end git and mercurial do not have any significant advantages over
each other feature wise.
Ok, then let's agree to disagree regarding this point.
You are at Sun, while I am just a community member. It's up to you and
your colleagues to take a final decision.
However, I see some advantages for mercurial regarding speed of
development, extensibility, customizability and learning curve.
My personal impression looking at mailing lists is that the Git crew is
larger and more active, than Mercurial's.
Regarding the extensibility, Mercurial has the extension concept though I
am not aware how powerful it really is and whether base functionality of
Mercurial can really be totally overridden. With Git you would have to
apply some patches. In both cases some work with sources has to be done
which is no real problem for a developer, and applied changes will look
like core functionality.
While the extension concept is clearer, I have more than once run into
trouble because the extension suddenly became incompatible with an updated
core version (of course the same can happen with a Git patch, just wanted
to point out that having an extension does not mean it will work forever).
Regarding learning curve, well there are Git wrappers, like 'Easy Git'
<http://www.gnome.org/~newren/eg> that remove that burden.
May I propose that you join the mercurial pilot? It will let you get to
know the tool and its a _lot_ less pain than using SVN in every possible
way.
I have no objections regarding taking a look at it. Do I need special
access rights for it?
Although I am no OpenOffice.org developer, I currently have an SVN mirror
here and based on that a local Git repo that is connected to the SVN trunk
just for playing around.
BTW: An ideal DVCS in my mind would be a combination of Git's features and
speed, Mercurial's UI and Bazaars ability to track copies, moves and
directories.
Best regards
Guido
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
