rich-root vs subtree formats for 1.6

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Martin Pool wrote:
> At the moment we are introducing just these two new stable stacking
> pack format names in 1.6:
> 
> format_registry.register_metadir('1.6',
>     'bzrlib.repofmt.pack_repo.RepositoryFormatKnitPack5',
>     help='A branch and pack based repository that supports stacking. ',
>     branch_format='bzrlib.branch.BzrBranchFormat7',
>     tree_format='bzrlib.workingtree.WorkingTreeFormat4',
>     )
> format_registry.register_metadir('1.6-rich-root',
>     'bzrlib.repofmt.pack_repo.RepositoryFormatKnitPack5RichRoot',
>     help='A branch and pack based repository that supports stacking '
>          'and rich root data (needed for bzr-svn). ',
>     branch_format='bzrlib.branch.BzrBranchFormat7',
>     tree_format='bzrlib.workingtree.WorkingTreeFormat4',
>     )
> 
> There is no 1.6-subtree, and neither of these formats have
> supports_tree_reference turned on.  I think this is what we want, as
> subtree is still experimental and there are experimental formats (like
> development1-subtree) that do have it.  Is that correct?
> 

I think so. At least, I would say if someone wants to experiment with -subtree
not having stacking isn't a big penalty. (And technically, they still could
with a custom layout, as stacking is a Branch thing, not a repo thing.)

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIkyONJdeBCYSNAAMRAmb9AJwJYtfdSkxaXbKTWIVsmco2I0gbogCePZNT
LNYhRPfT9/nAPUI2XHOYHxc=
=r55U
-----END PGP SIGNATURE-----