Re: Handling large files with GIT

From: Ben Clifford <benc@hawaga.org.uk>
Date: Feb 13, 2006, 01:26 UTC
Message-ID: <Pine.OSX.4.64.0602131416530.25089@piva.hawaga.org.uk>
In-Reply-To: <46a038f90602081435x49e53a1cgdc56040a19768adb@mail.gmail.com>

On Thu, 9 Feb 2006, Martin Langhoff wrote:

> I did suggest maildir,  where GIT is bound to do well as the content
> of the emails doesn't change but they just move around a lot. Though
> yes, trees are going to be nasty.

I've been keeping maildir in git for a few months, with mail being delivered into a git repo on one (permanently connected) host and me merging that branch into a repo on my laptop for reading (the intention being that I should be able to sync it back to the permanently connected host as I sometimes read mail there.

Alas, the merge part of this absolutely sucks -- as time goes by, its getting slower and slower (its taking an hour or so to do the merge, which has got to the point of being barely usable -- if it wasn't for the neat hack-value, I'd have given up on this by now).

I haven't really probed whats happening when I'm doing the merges in any depth, but I see a lot of index manipulation happening (git update-index, I think) to add and remove files where each invocation of that seems to be taking almost a whole second.

I wonder if the present merge algorithms perform especially badly in the case of a large number of files with lots of renames (and so lots of adds/removes) but no content changes? The merge should be able to happen entirely in the index, I think.

Perhaps one way to proceed would be for me to write a move-optimised merge strategy where I flip the index round and instead of saying "how has the content inside this filename changed?" I instead say "how has the filename associated with this content <hash> changed?"

A special-case on top of a move-optimised merge might be some maildir-aware filename handling that knows how to resolve conflicts when a particular content-hash has been renamed to two different names (eg. when one flag is added to a message in one repo and a different flag is added to a message in another repo).

Any advice/thoughts/suggestions-that-this-is-a-stupid-thing-to-do would be greatly appreciated.