Re: [REVISED PATCH 2/6] Introduce commit notes
- From
Shawn O. Pearce <spearce@spearce.org>
- Date
- Jul 20, 2007, 04:59 UTC
- Message-ID
- <20070720045909.GM32566@spearce.org>
- In-Reply-To
- <7vodi83fg7.fsf@assigned-by-dhcp.cox.net>
Junio C Hamano <gitster@pobox.com> wrote:
Show 8 quoted lines
> I've long time ago concluded that if we care about reliability > (and we do very much), a bisectable tree without breaking > backward compatibility is impossible. I was hoping to find a > "hole" in tree object format so that I can place an extended > section that is invisible to older versions of git, and place a > table that records offsets of each tree entries to help > bisection and/or perhaps a hash table to help look-up, but I do > not think it is possible.
...
> But the tree object format > is designed so tight that I do not see there is any place to put > an extension section.
I came to the same conclusion the last time I thought about this problem, for all the same reasons you outlined. And came up with pack v4. Because the only way I could see that we could produce a more optimal tree was to just use a different *compression* of the tree, while still keeping its data the same. Nico seemed to agree at the time, because he worked on the prototype with me. :-)
Its still hanging around in my fastimport repository. But has not been merged with any recent Git, and it still needs a lot of work.
-- Shawn.