git-sharp, was Re: Implementations of Git (was: re-licensing pkt-line.c code)

From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Sep 26, 2008, 11:42 UTC
Message-ID: <alpine.DEB.1.00.0809261335240.13830@pacific.mpi-cbg.de.mpi-cbg.de>
In-Reply-To: <m3od2c2k29.fsf@localhost.localdomain>

Hi,

On Thu, 25 Sep 2008, Jakub Narebski wrote:

> P.S. Do you know what is the status of git#, implementation of Git in 
> C#, which was one of this year Google Summer of Code projects?

As I suspected (hoping I would be wrong), the lack of communication (this time with the Git community) did not help at all.

The latest commit is this:

http://code.google.com/p/mono-soc-2008/source/detail?r=471

implementing a diff algorithm.

>From what I can tell from the source (I do not have a working C#

environment), the implementation repeats a lot of what Shawn told us earlier were his lessons from his first stab at JGit. For example, I saw that the contents of blobs are read into memory when the blob is read. I haven't checked closely, but I did not have the impression that there were accessors making it easy to turn this into a lazy read.

Take-home message: if you think you should develop Open Source in a secretive manner, you are almost certainly wrong, and your project will almost certainly be slow or even fail (BTW I also see exactly this same problem with Govind's try last year, which had a cool name, and then another cool name, but not much to show for it).

Ciao, Dscho