Re: [Monotone-devel] basic_io inventory (was xmtn: Emacs integration via DVC)

[Derek Scherger]

Thomas Keller wrote:
> Derek Scherger schrieb:
>> I don't think this is correct. Listing either old or new name of a
>> renamed path will include the *node* in a restriction and including the
>> node will list both old and new names in the resulting inventory output.
> 
> I've actually tested this out before and it is correct if a rename
> happens in the same directory. But, AFAIR, if you move / rename a file
> to a different directory you do _not_ get the renamed node outputted as
> well. Maybe we should fix that then?

I had a quick poke at this and can't seem to reproduce what you're
talking about. Can you come up with an example?

I did notice something odd with the way inventory handles restrictions
though. If seems that listing the old name of a renamed file fails with
inventory but not with status. Changing inventory to pass the parent_map
rather than the first roster in the parent_map when it creates its
node_restriction seems to fix it but this seems a bit strange. There's
only one parent so it would seem that they should be equivalent.

Zack, any ideas on this, the parent_map stuff is related to the
workspace merge changes iirc.

Cheers,
Derek