Proposals/feedback for a distributed version control system for cutecash?

[Christian Stimming]

Am Dienstag, 6. April 2010 schrieb John Ralls:
> On Apr 6, 2010, at 2:28 AM, Colin Law wrote:
> > 2010/4/6 Christian Stimming <stimming at tuhh.de>:
> >> In particular, downloading everything by "git clone" doesn't take that
> >> long, hence I don't understand some comments here which mentioned the
> >> need for an extra download-all-in-one-file requirement.
> >
> > It is about 90MB but github sourced it at my full connection speed of
> > 2Mb (in the UK) and it took less than 9 minutes go clone.  Since
> > cloning is a rare occurrence for each user I do not see that as a
> > problem, for most anyway.
> 
> It doesn't take long to clone because it doesn't have any history: It was a
>  "flat" import from svn (i.e., git svn fetch
>  http://svn.gnucash.org/gnucash/trunk instead of
>  http://svn.gnucash.org/gnucash).

I can't follow you here. As I said, I just forked from timabell's SVN import, 
and he indeed converted everything including trunk and 2.2 and 2.whatnot 
branches and tags until last April. This also gets cloned into the local git 
repo.

I agree my update of timabell's clone (i.e. the SVN commits from last April 
until now) are indeed just the commits from trunk, without any tags or other 
branches. But all the data from before that is in there.

>  The actual migrated repo with its 20
>  years of changes will be *much* larger.

No, the actual migrated repo (with the first commit from linas on 1997-11-01) 
does not take longer when accessed through git. The one we're currently 
talking about is already almost everything.

And for the record, the download ("git clone") took on the order of 2 minutes 
here (at a 16MBit tube). This, together with Colin's number above, rather 
tells me "git clone" is as efficient as it could get, as only the download 
speed is the bottleneck anymore.

Regards,

Christian