[MERGE] Describe a hash trie based inventory

[Robert Collins]

On Tue, 2008-08-26 at 09:45 +0200, Vincent Ladeuil wrote:
> 
> Have you considered handling the inventory as a flat file-ids list
> instead of a path tree ?

Well, essentially this does treat it as a flat list of file ids - the
list is stored via the hash trie, with the file id as the key.

> This could keep the nice properties of hash tries while
> addressing the 10.000 files in one directory corner case as well
> as the 'Goal satisfaction point 4' above.

The only issue with point 4 is the multiple lookups - to solve that you
need to have the full path stored in some fashion, and that implies
recursively altering every entry when its parent is renamed even if the
entry itself isn't altered. The hash trie approach doesn't know about
child paths, so its not amenable to doing that without doing a scan.

Thanks for the review.

-Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20080826/1ad9f38b/attachment.pgp