[MERGE] New 'development' format.

[Aaron Bentley]

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

Robert Collins wrote:
> On Mon, 2008-01-07 at 00:06 -0500, Aaron Bentley wrote:
>> I feel that non-developers should not be playing with experimental
>> formats until they are proposed for becoming stable.
>
> Well, this leads to a serious chicken and egg problem. packs were stable
> in terms of disk layout and data many weeks before they were proposed
> for being non-experimental, and yet I only know of 3-4 beta testers in
> that entire time. We had similar problems with the introduction of
> weaves in the first place, and again at the split out metadir work. I
> consider this a process bug, which this patch is aiming to address.

Well, I don't know what to suggest.  I can't imagine why anyone would
trust their real data, not test data, to a format that wasn't stable.  I
want to hear "I think this is ready to become stable" first.

We have plenty of data that can be *imported*, which helps somewhat.  We
have lots of tests, which also help.  Once you're announce that it's
ready, I'll commit to it.  But during packs development, every time you
announced "the format has changed-- here's how to upgrade", I was
relieved that I was not using packs yet.

> Thats fine; I think its reasonable to make a decision about whether you
> want to allow pulling to/from formats you are working on. Note that if
> you allow fetching from your new format

I will not be working on a new format, because no one has approved my
initial work.  Since it looks unlikely that my initial work will be
merged, it would be pointless to continue.

>>> +If you are using ``bzr-svn`` or are testing the prototype subtree support,
>>> +you can still use and assist in testing the development formats. The
>>> +commands to use are identical to the ones given above except that the
>>> +name of the format to use is ``development-subtree``.
>> I am also uncomfortable with this.  If you want bzr-svn users doing
>> testing, then please provide a rich-root variant of development.
>
> I don't particularly care about bzr-svn users testing the stuff I'm
> working on, but we can do this. The main problem though is that 'bzr
> help formats' gets quite long :(.

Well, it is your choice how you handle it, but I do not want to be
encouraging bzr-svn users to be using subtree, because it is supposed to
be incompatible with rich-root.  I don't understand why you can pull
from subtree to rich-root, but I consider it a bug.

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

iD8DBQFHgdBi0F+nu1YWqI0RAuOjAJ9ukXJd2bLCy9l6MYORAU4XBN1YWQCZAbGR
rLCn0JDqfr1j3ajQBkFczH0=
=v6nr
-----END PGP SIGNATURE-----