At http://people.ubuntu.com/~robertc/baz2.0/inventory
------------------------------------------------------------ revno: 3644 revision-id: [EMAIL PROTECTED] parent: [EMAIL PROTECTED] committer: Robert Collins <[EMAIL PROTECTED]> branch nick: inventory timestamp: Wed 2008-09-24 17:12:24 +1000 message: Define CHK and very minor tweaks to the inventory text. modified: doc/developers/inventory.txt inventory.txt-20080103013957-opkrhxy6lmywmx4i-1 === modified file 'doc/developers/inventory.txt' --- a/doc/developers/inventory.txt 2008-08-28 03:37:52 +0000 +++ b/doc/developers/inventory.txt 2008-09-24 07:12:24 +0000 @@ -230,7 +230,9 @@ Canonical form -------------- -There are three fragment types for the canonical form: +There are three fragment types for the canonical form. Each fragment is +addressed using a Content Hash Key (CHK) - for instance +"sha1:12345678901234567890". root_node: (Perhaps this should be inlined into the revision object). HASH_INVENTORY_SIGNATURE @@ -264,7 +266,6 @@ start with just a single leaf node with an empty prefix. - Apply ----- @@ -290,10 +291,11 @@ To generate the list of altered file ids, we do an entry by entry comparison of the full contents of every leaf node that the two inventories do not have in common. To do this, we start at the root node, and follow every CHK pointer -that is only in one tree. We can then bring in all the value from the leaf +that is only in one tree. We can then bring in all the values from the leaf nodes and do a set difference to get the altered ones, which we would then parse. + Radix tree based inventories ============================ -- bazaar-commits mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/bazaar-commits
