Re: [jgit-dev] FYI, history rewrite coming

["Shawn O. Pearce" <spearce@xxxxxxxxxxx>]

Robin Rosenberg <robin.rosenberg@xxxxxxxxxx> wrote:
> 
> Why do we have to be history revisionists? Can't you
> just start a new history and tell people how to graft
> the real history. We did that once. 

I could.

But then I might need to redo the entire initial contribution to be
as-of our current state of tree.  Which means regathering the initial
contribution agreements, at least from people who have contributed
since the initial contribution and who aren't project committers.
And possibly also for project committers who have contributed only
after the initial contribution.

If you want to try going down that road, you're welcome to do it.
I just think its going to be a great deal of trouble.

> Don't you tell me the CVS/SVN based projects rewrite their
> history.

The CVS/SVN projects rewrite their history at Eclipse.org.

In CVS, incorrect files like ours currently are, just get rm -f'd
on the server by webmaster.  And someone checks in a fixed copy
and the project pretends to move on.  They consider it a feature
that if you update after the rm on the server, that cvs update will
also rm the file from your harddrive.  The Eclipse IP team likes
the idea of a remote wipe of source code.  *shrug*

The SVN projects, they dump their repository and reimport anytime
something like this happens.  Yes.

Its a risk we take by doing development through the parallel
IP process.  AFAIK there are only two ways to avoid this pain:

1) Abandon Eclipse.org and develop the project externally.
   Import releases into Orbit each time we need it.

2) Always wait for a full contribution review to be done before
   development can continue.  That means we would still be waiting
   to get any work done on JGit, since the IP team still hasn't
   completed their _initial_ content review.

-- 
Shawn.