Re: [Gnu-arch-users] Re: [BUG] feature plan -- Selected inventory

[Jason McCarty]

Tom Lord wrote:
> Suppose that I have a selection specification that picks out a subset
> of the ORIG tree and my MOD tree.  Sure, I could `inventory' keeping
> that selection spec in mind and, if the selection spec doesn't have
> any id-based rules, avoid inventorying all of ORIG or MOD.
> 
> But now if I'm computing a changeset:  files may appear in the
> selected part of MOD that were absent in ORIG or they may be ORIG but
> not in MOD.   Are those adds and deletes?  Or are they renames from
> and to some non-selected part of the tree?   The only way to tell is
> to inventory all of ORIG and all of MOD.

But we can easily detect when a full inventory is necessary and only do
it if needed, no? Isn't it correct that a full-tree inventory is only
necessary when we detect an out-of-selection change? That is, an id-tag
enters or exits the selection space, through an add, delete, or rename
within the selection.  

-- 
Jason McCarty <address@hidden>