[Kde-scm-interest] Have we arrived to a dead end?

[Ian Monroe]

On Fri, Feb 12, 2010 at 6:44 AM, Robert Wohlrab <robert.wohlrab at gmx.de> wrote:
> Ian Monroe wrote:
>> On Fri, Feb 12, 2010 at 6:35 AM, Robert Wohlrab <robert.wohlrab at gmx.de>
> wrote:
>> > Riccardo Iaconelli wrote:
>> >> So, what are we doing now? Giving up on git and telling people to just
>> >> screw up and use git-svn? Think more? Hire someone to make a better
>> >> git? Or what?
>> >
>> > No, just implement the narrow transfer protocol and handling. Or do you
>> > just give up KDE because kget doesn't support freenet? I would do and
>> > change to google os.
>> >
>> > http://www.thousandparsec.net/~tim/media+git.pdf
>> > http://git.wiki.kernel.org/index.php/SoC2009Ideas#Narrow_and_Sparse_clone
>> >_support
>>
>> Again this narrow clone support is purely for consumers of the git
>> repo. If you wanted to commit anything you'd still need the whole
>> repo.
>
> Who says that? You have _____aalllllll____ metadata. Only the blobs are
> missing partially. So you can completely calculate the sha for your trees and
> all your parent trees. And this means that you can also create commits and
> tags and so on. All you need is there. You must only calculate new sha sums
> for the stuff you changed... and you know that you only change stuff in your
> narrow clone. This also means that you can push and pull. You can even look at
> the logs. Only accessing a tree with missing blobs isn't possible.

I was just going off of the GSoC description page.