Merge lp:~lifeless/bzr/bug-398668 into lp:~bzr/bzr/trunk-old

The diff is likely still a problem - here is a manual diff.

=== modified file 'NEWS'
--- NEWS 2009-08-15 09:15:03 +0000
+++ NEWS 2009-08-17 23:19:17 +0000
@@ -9,6 +9,21 @@
 In Development
 ##############

+Compatibility Breaks
+********************
+
+* The default format for bzr is now ``2a``. This format brings many
+ significant performance and size improvements. bzr can pull from
+ any existing repository into a ``2a`` one, but can only transfer
+ into ``rich-root`` repositories from ``2a``. The Upgrade guide
+ has more information about this change. (Robert Collins)
+
+New Features
+************
+
+Bug Fixes
+*********
+
 Improvements
 ************

@@ -16,6 +31,18 @@
   --version`` and ``bzr selftest``.
   (Martin Pool, #409137)

+Documentation
+*************
+
+API Changes
+***********
+
+Internals
+*********
+
+Testing
+*******
+
 bzr 1.18
 ########

@@ -58,6 +85,11 @@
 * StreamSource generates rich roots from non-rich root sources correctly
   now. (Andrew Bennetts, #368921)

+* When deciding whether a repository was compatible for upgrading or
+ fetching, we previously incorrectly checked the default repository
+ format for the bzrdir format, rather than the format that was actually
+ present on disk. (Martin Pool, #408824)
+
 Improvements
 ************

@@ -105,8 +137,8 @@
   can run the https tests. (Denys Duchier, #392401)

-bzr 1.18
-########
+bzr 1.18rc1
+###########

 :Codename: little traveller
 :1.18rc1: 2009-08-10

=== modified file 'bzrlib/branch.py'
--- bzrlib/branch.py 2009-08-17 06:17:21 +0000
+++ bzrlib/branch.py 2009-08-17 06:19:20 +0000
@@ -2067,7 +2067,7 @@
 BranchFormat.register_format(__format6)
 BranchFormat.register_format(__format7)
 BranchFormat.register_format(__format8)
-BranchFormat.set_default_format(__format6)
+BranchFormat.set_default_format(__format7)
 _legacy_formats = [BzrBranchFormat4(),
     ]
 network_format_registry.register(

=== modified file 'bzrlib/bzrdir.py'
--- bzrlib/bzrdir.py 2009-08-17 06:17:21 +0000
+++ bzrlib/bzrdir.py 2009-08-17 07:47:50 +0000
@@ -130,8 +130,7 @@

     def check_conversion_target(self, target_format):
         target_repo_format = target_format.repository_format
- source_repo_format = self._format.repository_format
- source_repo_format.check_conversion_target(target_repo_format)
+ self.open_repository()._format.check_conversion_target(target_repo_format)

     @staticmethod
     def _check_supported(format, allow_unsupported,
@@ -3865,11 +3864,11 @@
 # The following format should be an alias for the rich root equivalent
 # of the default format
 format_registry.register_metadir('default-rich-root',
- 'bzrlib.repofmt.pack_repo.RepositoryFormatKnitPack4',
- help='Default format, rich root variant. (needed for bzr-svn and bzr-git).',
- branch_format='bzrlib.branch.BzrBranchFormat6',
- tree_format='bzrlib.workingtree.WorkingTreeFormat4',
+ 'bzrlib.repofmt.groupcompress_repo.RepositoryFormat2a',
+ branch_format='bzrlib.branch.BzrBranchFormat7',
+ tree_format='bzrlib.workingtree.WorkingTreeFormat6',
     alias=True,
- )
+ help='Same as 2a.')
+
 # The current format that is made on 'bzr init'.
-format_re...

There are still two sub-proposals up for review.
https://code.edge.launchpad.net/~lifeless/bzr/test-log/+merge/10137
https://code.edge.launchpad.net/~lifeless/bzr/CHKRepo._iter_inventory_xmls/+merge/10132

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

...

> # The following format should be an alias for the rich root equivalent
> # of the default format
> format_registry.register_metadir('default-rich-root',
> - 'bzrlib.repofmt.pack_repo.RepositoryFormatKnitPack4',
> - help='Default format, rich root variant. (needed for bzr-svn and bzr-git).',
> - branch_format='bzrlib.branch.BzrBranchFormat6',
> - tree_format='bzrlib.workingtree.WorkingTreeFormat4',
> + 'bzrlib.repofmt.groupcompress_repo.RepositoryFormat2a',
> + branch_format='bzrlib.branch.BzrBranchFormat7',
> + tree_format='bzrlib.workingtree.WorkingTreeFormat6',
> alias=True,
> - )
> + help='Same as 2a.')
> +
> # The current format that is made on 'bzr init'.
> -format_registry.set_default('pack-0.92')
> +format_registry.set_default('2a')

^- I'm 75% sure that we want the default WT format to be
"WorkingTreeFormat5" and not 6. 5 includes support for filters, 6
brings in support for 'views'. And View support was considered
incomplete, and thus why it was split out into a different format.

...

v- Obviously changing the above to '5' will bump all of these as well.

> Format:
> control: Meta directory format 1
> - working tree: Working tree format 4
> + working tree: Working tree format 6
> branch: Branch format 4
> repository: Weave repository format 6
>
> @@ -453,7 +449,7 @@
>
...

> -__default_format = WorkingTreeFormat4()
> +__default_format = WorkingTreeFormat6()
> WorkingTreeFormat.register_format(__default_format)
> -WorkingTreeFormat.register_format(WorkingTreeFormat6())
> WorkingTreeFormat.register_format(WorkingTreeFormat5())
> +WorkingTreeFormat.register_format(WorkingTreeFormat4())
> WorkingTreeFormat.register_format(WorkingTreeFormat3())
> WorkingTreeFormat.set_default_format(__default_format)
> # formats which have no format string are not discoverable

^- This would also be effected.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkqKrzsACgkQJdeBCYSNAAONjACgnQBU95OlegGgtPzeaPVesvxO
/p0AnR+zohxUXfeHnv3QfQ3Q4RT+tcAc
=3UXs
-----END PGP SIGNATURE-----

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

Robert Collins wrote:
> Robert Collins has proposed merging lp:~lifeless/bzr/bug-398668 into lp:bzr.
>
> The diff is likely still a problem - here is a manual diff.

Sorry, I meant to include:
  review: approve

Provided there is agreement about WT5 vs WT6.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkqKr1MACgkQJdeBCYSNAAMrAwCfSQLQ5ykqfetCyw2Fdp2YqGDT
NP4AnRP63XK5SOcngerDN+SdwjWUFc+l
=cqKQ
-----END PGP SIGNATURE-----

On Tue, 2009-08-18 at 13:42 +0000, John A Meinel wrote:
> Review: Approve
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Robert Collins wrote:
> > Robert Collins has proposed merging lp:~lifeless/bzr/bug-398668 into lp:bzr.
> >
> > The diff is likely still a problem - here is a manual diff.
>
> Sorry, I meant to include:
> review: approve
>
> Provided there is agreement about WT5 vs WT6.

2a is format 6, and thats a supported format. Assuming that you're
right, we need to actually downgrade 2a in 1.16/1.17/1.18 as well -
definitely out of scope for this patch - and while changing the default
will increase the number of 2a users out there, we have from now until
the release to decide on this.

There is one outstanding log test I need to change as well, but I'll
land that change asynchronously - I don't want to let this [somewhat
fragile branch] bitrot at all.

Thanks for the review.

-Rob

Robert Collins wrote:
> On Tue, 2009-08-18 at 13:42 +0000, John A Meinel wrote:
>> Review: Approve
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Robert Collins wrote:
>>> Robert Collins has proposed merging lp:~lifeless/bzr/bug-398668 into lp:bzr.
>>>
>>> The diff is likely still a problem - here is a manual diff.
>> Sorry, I meant to include:
>> review: approve
>>
>> Provided there is agreement about WT5 vs WT6.
>
> 2a is format 6, and thats a supported format. Assuming that you're
> right, we need to actually downgrade 2a in 1.16/1.17/1.18 as well -

I'd prefer to keep the format exactly as it is now and stay with WT6.
I'm happy to make filtered views an experimental-only feature for 2.0,
say so loudly in the doc and hide the view command if appropriate. But I
don't want to remove the ability to store the filtered view metadata in
the default format.

Ian C.

On Wed, 2009-08-19 at 01:03 +0000, Ian Clatworthy wrote:
>
> > 2a is format 6, and thats a supported format. Assuming that you're
> > right, we need to actually downgrade 2a in 1.16/1.17/1.18 as well -
>
> I'd prefer to keep the format exactly as it is now and stay with WT6.
> I'm happy to make filtered views an experimental-only feature for 2.0,
> say so loudly in the doc and hide the view command if appropriate. But
> I
> don't want to remove the ability to store the filtered view metadata
> in
> the default format.

I don't have any particular opinion.

The default is 2a with wt6 trees, at this point.

Cheers,
Rob