Re: NYU Open Source Programming Class Releases JavaGit API 0.1.0 Alpha

From: Shawn O. Pearce <spearce@spearce.org>
Date: Aug 5, 2008, 01:20 UTC
Message-ID: <20080805012007.GB32543@spearce.org>
In-Reply-To: <20080805011051.GC32184@machine.or.cz>

Petr Baudis <pasky@suse.cz> wrote:

Show 15 quoted lines

> On Tue, Aug 05, 2008 at 01:36:15AM +0200, Johannes Schindelin wrote:
> > And I absolutely share the puzzlement of Shawn why anybody would try to do 
> > their own thing in Java, instead of contributing to jgit.
> > 
> > Especially the fact that "JavaGit" seems to repeat the all-to-obvious 
> > error of Eclipse: to make a platform-independent library that depends on 
> > platform dependent components.
> 
> [Totally naive question coming with no knowledge of actual size of
> current JGit's feature set:]
> 
> Wouldn't it make sense to have an ability to fall back on Git calls for
> functionality that is not implemented natively yet? If you then don't
> have Git available, you just don't get the functionality. What's the
> catch?

Someone else asked to do this in JGit before (it was privately off-list to Robin and myself). Both of our responses were that we were not that interested in having such dependencies in the main JGit tree at this time, but that JGit is BSD and nothing would stop them from forking our tree and trying it out. If there really was value there we would have to consider merging it back.

Nothing came of it as far as I know.

Someone recently ported git-blame into JGit, and it looks like they did a good job, but it was a straight C->Java port. Since JGit is licensed under the BSD and builtin-blame.c is under the GPL we couldn't accept it as-is. That side project produced a lot more results than anything else of late, but the GPL virual clause kicked in and prevented direct inclusion.

-- 
Shawn.