[MERGE][1.6.1][bug #262333] New repo format for rich roots with a proper serializer

[Matt Nordhoff]

John Arbash Meinel wrote:
> Aaron uncovered that --1.6-rich-root was actually a bit confused as to whether
> it was trying to be --1.6-subtree or not. It seems the serializer supported
> subtrees, but the layering did not. So you couldn't directly stack on a
> --rich-root-pack repository, and you might accidentally fetch from a
> --dirstate-with-subtree one.
> 
> So this patch deprecates the --1.6-rich-root format, in favor of
> --1.6.1-rich-root.
> 
> I'm not 100% sure about the naming. I could name it --1.6-rich-root and
> --1.6-old-rich-root. (or --1.6-rich-root-broken).
> 
> I'd like to get some form of this into 1.6.1, so it would be nice to get feedback.
> 
> John
> =:->

I'm not a reviewer, but...

bb:tweak/bb:comment

I'm +0.9 on calling the new format "--1.6-rich-root" and renaming the
old format to "--1.6-rich-root-broken" or something similar. This is a
serious issue, and there's historical precedent: the original pack
formats were renamed once, right?

Also, this is weak, but if the old format wasn't renamed, a user could
accidentally type in "--1.6-rich-root" when creating a branch or
something. At the very least, I think that's more likely to happen than
someone being confused by a one-week-old, non-default format changing names.

Also:

> === modified file 'bzrlib/bzrdir.py'
> --- bzrlib/bzrdir.py	2008-08-19 21:04:22 +0000
> +++ bzrlib/bzrdir.py	2008-08-29 02:44:39 +0000
> @@ -153,6 +153,7 @@
>              raise errors.UnsupportedFormatError(format=format)
>          if recommend_upgrade \
>              and getattr(format, 'upgrade_recommended', False):
> +            import pdb; pdb.set_trace()
>              ui.ui_factory.recommend_upgrade(
>                  format.get_format_description(),
>                  basedir)
> @@ -3025,13 +3026,23 @@

Whoops?
--