Why git instead of mercurial?

On 27/03/2008, Robert D. [email protected] wrote:

On Wed, Mar 26, 2008 at 9:54 PM, Yukihiro M. [email protected] wrote:

  • having explicit central repository is a good thing for Ruby

Sorry to be bold for once Matz, but this sentence surprises me a lot,
why is a single point of failure a good thing?

You can make as many backups as you like, there’s nothing stopping you.

What I like most about Mercurial (and distributed VCS in general) to
have my repository on 4 different PCs and 3 pendrives, while top is
not on all seven it is almost impossible to really lose data.

The single central repository is a workflow pattern - the address
which you officially publish and from which people check out the
sources.

While CVS or SVN enforce this pattern distributed VCSs do not.

Thanks

Michal

Hi,

In message “Re: Why git instead of mercurial?”
on Thu, 27 Mar 2008 23:21:50 +0900, “Robert D.”
[email protected] writes:

|> * having explicit central repository is a good thing for Ruby
|Sorry to be bold for once Matz, but this sentence surprises me a lot,
|why is a single point of failure a good thing?

Unless we abandon central repository model, we need an authenticated
repository anyway. And I didn’t want to give up that clear and simple
model.

          matz.

On Thu, Mar 27, 2008 at 4:36 PM, Yukihiro M. [email protected]
wrote:

Unless we abandon central repository model, we need an authenticated
repository anyway. And I didn’t want to give up that clear and simple
model.

                                                    matz.

To Matz and Michael
I guess it is just a completely different way to work, but I guess I
frequently do things differently so that should not alarm you :slight_smile:
Just for your information Mercurial can use ssh authenticated
repositories so that would not be a problem at all.
I feel that sometimes people forget that a distributed CVS is also a
centralized one, just a matter of usage, right?

Cheers
Robert


http://ruby-smalltalk.blogspot.com/


Whereof one cannot speak, thereof one must be silent.
Ludwig Wittgenstein

On 27/03/2008, Robert D. [email protected] wrote:

To Matz and Michael
I guess it is just a completely different way to work, but I guess I
frequently do things differently so that should not alarm you :slight_smile:

I started to use a distributed VCS for my work, and I like it. I can
set up a repo and make changes anywhere, and later push them anywhere
else.

However, for a larger project which you publish for everybody to see
and check out a more centralized workflow is needed. Only one repo is
the published “official” one.

Just for your information Mercurial can use ssh authenticated
repositories so that would not be a problem at all.
I feel that sometimes people forget that a distributed CVS is also a
centralized one, just a matter of usage, right?

Sure, but when you are used to centralization, and want it anyway
there is no gain in distributed VCS.

Thanks

Michal

On 27 Mar 2008, at 14:21, Robert D. wrote:

On Wed, Mar 26, 2008 at 9:54 PM, Yukihiro M. <[email protected]

wrote:

  • having explicit central repository is a good thing for Ruby
    Sorry to be bold for once Matz, but this sentence surprises me a lot,
    why is a single point of failure a good thing?
    What I like most about Mercurial (and distributed VCS in general) to
    have my repository on 4 different PCs and 3 pendrives, while top is
    not on all seven it is almost impossible to really lose data.
    Never been afraid of this?

I have a copy in git, so does matz, and I’m sure there are more
people. These carry the full repo history, sufficient to rebuild an
svn repo, or just keep in git under such a situation.

I know of many people that have checkouts under svn.

I don’t think there’s a problem with potential loss for something this
widely used.