Merge lp:~jelmer/bzr-builddeb/no-upstream-branch into lp:bzr-builddeb

[Jelmer Vernooij]

This fixes a regression in one of my earlier patches - upstream_branch can actually be None.

[Robert Collins (lifeless)]

DoIt

[lp:~jelmer/bzr-builddeb/no-upstream-branch]

460. By James Westby on 2010-06-30

Merge fixes from I.

461. By James Westby on 2010-07-02

Improvements to merge-upstream help. Thanks Martin.

462. By KarlGoetz on 2010-07-07

Fix readme rstification from Karl Goetz

463. By James Westby on 2010-07-29

Fix errors on import of renamed files.

When a file was renamed we would erroneously assign the file id to two paths.
We avoid that by ensuring we take the one from the upstream branch if there is
one. Thanks Rob.

464. By James Westby on 2010-07-29

Fixed a logic error that stops -r working in merge-upstream.

465. By James Westby on 2010-07-30

Try importing from debian first. Thanks Jelmer.

466. By James Westby on 2010-08-11

Fix doctest failures with newer python-debian. Thanks Jelmer.

467. By James Westby on 2010-08-11

Determine Bazaar home directory using bzrlib to prevent test isolation issues.

468. By Jelmer Vernooij on 2010-08-11

Upload 2.5.

469. By James Westby on 2010-08-19

Don't fail if asked to use a .bz2 tarball that is already in the desired location.

470. By James Westby on 2010-08-19

Fix two more causes of debian_bundle deprecation warnings.

471. By James Westby on 2010-08-19

Avoid using UpstreamBranchSource if we don't know the upstream branch in merge mode.

472. By James Westby on 2010-08-19

Avoid an error on merge-upstream with an unrelated branch. Thanks Andrew.

473. By James Westby on 2010-08-19

Add changelog entries for the last couple of fixes.

474. By James Westby on 2010-08-19

Don't strip -n from the version we get in merge-upstream.

Some upstreams have a dash in there, and trying to support both means
supporting both badly. If you are used to doing

  bzr merge-upstream --version <package version>

then it will no longer work for you, use the upstream version instead.

475. By James Westby on 2010-08-20

Refactor cmd_merge_upstream in to smaller chunks.

476. By James Westby on 2010-09-09

Avoid duplicate file ids in the import code with renames.

Don't crash when doing merge-upstream with a branch that does a rename
and then ships another file with the old path in the tarball that isn't
in the branch.

We would assign the same id to both paths, which obviously doesn't work.

Instead we now do a first pass to determine which file ids in the tree
will be assigned to different paths, and then avoid re-using that file
id with the old path.

477. By James Westby on 2010-09-10

Split file_ids_from in to one we should take renames from, and others.

Having multiple trees that we may follow renames from led to lots of
bugs, and turned out to be unnecessary.

Split file_ids_from to give a single tree that will be authoritative,
and a list of others that we will take file ids from, but only if we would
otherwise generate one.

478. By James Westby on 2010-09-10

Merge Jelmer's branch and credit him with the fix.

=== modified file 'cmds.py'
--- cmds.py	2010-05-28 20:49:11 +0000
+++ cmds.py	2010-06-15 18:32:27 +0000
@@ -380,8 +380,10 @@
                 upstream_branch, upstream_revision = self._get_upstream_branch(
                     merge, export_upstream, export_upstream_revision, config,
                     changelog.version)
-                upstream_sources.append(UpstreamBranchSource(upstream_branch,
-                    {changelog.version.upstream_version: upstream_revision}))
+                if upstream_branch is not None:
+                    upstream_sources.append(UpstreamBranchSource(
+                      upstream_branch, {
+                        changelog.version.upstream_version: upstream_revision}))
             elif not native and config.upstream_branch:
                 upstream_branch = Branch.open(config.upstream_branch)
                 upstream_sources.append(UpstreamBranchSource(upstream_branch))