Re: Voting: do you want to move to git?

[Dr. H. Nikolaus Schaller]

Am 14.02.2012 um 15:06 schrieb Thomas Davie:

> 
> On 14 Feb 2012, at 13:48, David Chisnall wrote:
> 
>> On 14 Feb 2012, at 13:46, Gregory Casamento wrote:
>> 
>>> Keeping SVN is a mistake given what has been demonstrated by GNA's
>>> failure.   SVN gives us a single point of failure.   A DVCS would
>>> eliminate that problem.
>> 
>> I'm not sure I fully agree with this.  The single point of failure exists in 
>> any system where w have an authoritative repository.  We could have an 
>> official git-svn mirror somewhere, or even a svn mirror using svncync that 
>> became the authoritative one when GNA was down if the only issue is being 
>> able to get at the code when gna is down.
> 
> Just to throw an extra spanner in the discussion – darcs (the original DVCS) 
> solves the above problem by regarding repositories as sets of patches instead 
> of as lists of states, which allows you to take patches from any repository 
> and push them into any other repository (given some constraints).  This means 
> you can do away with the idea of an "authoritative" repo, and happily push 
> changes between each other without issue.

Hm.
Do we want a rock solid system to work with (hint: there are enough bugs and 
omissions in the code) or do we want to turn GNUstep an experiment for 
alternate VCS (which may reveal unknown bugs)?

My philosophy: use the most mature tool you can get. This is rarely the latest 
one but also not the oldest one. It is often the most widely used one.
Then, you can concentrate on your own problems and don't have to fix the tools 
every now and then.

Nikolaus