Should be able to create colocated branches with "/" in the name to allow for hierarchical branch names

[Neil Martinsen-Burrell]

bzr-colo has had the ability to create branch names such as "origin/trunk" or "bugs/123456". This allows one to impose some order on the possibly many named branches. It would be nice if this worked in the core colocated workspace support. Something like

$ bzr init --development-colo file:test,branch=trunk
$ cd test
$ bzr branch file:,branch=origin/trunk

which currently fails with

nmb@garble[~/tmp]$ bzr init --development-colo file:test,branch=trunk
Created a lightweight checkout (format: development-colo)
nmb@garble[~/tmp]$ cd test
nmb@garble[~/tmp/test]$ bzr -Derror branch file:,branch=origin/trunk
bzr: ERROR: bzrlib.errors.NotBranchError: Not a branch: "/Users/nmb/tmp/test/,branch=origin/trunk/".

Traceback (most recent call last):
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/commands.py", line 919, in exception_to_return_code
    return the_callable(*args, **kwargs)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/commands.py", line 1124, in run_bzr
    ret = run(*run_argv)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/commands.py", line 672, in run_argv_aliases
    return self.run(**all_cmd_args)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/commands.py", line 694, in run
    return self._operation.run_simple(*args, **kwargs)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/cleanup.py", line 136, in run_simple
    self.cleanups, self.func, *args, **kwargs)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/cleanup.py", line 166, in _do_with_cleanups
    result = func(*args, **kwargs)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/builtins.py", line 1332, in run
    from_location)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/controldir.py", line 764, in open_tree_or_branch
    controldir = klass.open(location)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/controldir.py", line 673, in open
    return klass.open_from_transport(t, _unsupported=_unsupported)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/controldir.py", line 702, in open_from_transport
    find_format, transport, redirected)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/transport/__init__.py", line 1701, in do_catching_redirections
    return action(transport)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/controldir.py", line 690, in find_format
    transport, _server_formats=_server_formats)
  File "/Users/nmb/src/bzr/bzr.dev/bzrlib/controldir.py", line 1135, in find_format
    raise errors.NotBranchError(path=transport.base)
NotBranchError: Not a branch: "/Users/nmb/tmp/test/,branch=origin/trunk/".

[Jelmer Vernooij (jelmer)]

For ambiguity rasons, this would have to be file:,branch=origin%2Ftrunk rather than file:,branch=origin/trunk.

Other than that it seems like a good idea to me.

[Alexander Belchenko (bialix)]

Jelmer Vernooij пишет:
> For ambiguity rasons, this would have to be file:,branch=origin%2Ftrunk
> rather than file:,branch=origin/trunk.

Why so? That %2F is OK for GUI users, in which case GUI tool will
substitute / to %2F under the hood, but that's bad for CLI users.
Non-intuitive at all.

[Jelmer Vernooij (jelmer)]

Am 23/12/11 13:05, schrieb Alexander Belchenko:
> Jelmer Vernooij пишет:
>> For ambiguity rasons, this would have to be file:,branch=origin%2Ftrunk
>> rather than file:,branch=origin/trunk.
> Why so? That %2F is OK for GUI users, in which case GUI tool will
> substitute / to %2F under the hood, but that's bad for CLI users.
> Non-intuitive at all.

Compare with file://foo,branch=foo,a/file/in/the/branch

Cheers,

Jelmer

[Alexander Belchenko (bialix)]

Jelmer Vernooij пишет:
> Am 23/12/11 13:05, schrieb Alexander Belchenko:
>> Jelmer Vernooij пишет:
>>> For ambiguity rasons, this would have to be file:,branch=origin%2Ftrunk
>>> rather than file:,branch=origin/trunk.
>> Why so? That %2F is OK for GUI users, in which case GUI tool will
>> substitute / to %2F under the hood, but that's bad for CLI users.
>> Non-intuitive at all.
>
> Compare with file://foo,branch=foo,a/file/in/the/branch

I've compared. You have a comma after foo.
What about:

file://foo,branch=foo/bar,a/file/in/the/branch

What's wrong here?

[Jelmer Vernooij (jelmer)]

Am 23/12/11 13:50, schrieb Alexander Belchenko:
> Jelmer Vernooij пишет:
>> Am 23/12/11 13:05, schrieb Alexander Belchenko:
>>> Jelmer Vernooij пишет:
>>>> For ambiguity rasons, this would have to be file:,branch=origin%2Ftrunk
>>>> rather than file:,branch=origin/trunk.
>>> Why so? That %2F is OK for GUI users, in which case GUI tool will
>>> substitute / to %2F under the hood, but that's bad for CLI users.
>>> Non-intuitive at all.
>> Compare with file://foo,branch=foo,a/file/in/the/branch
> I've compared. You have a comma after foo.
Sorry, that actually should have been
file://foo/bar/blie,branch=abranch/file/in/the/branch

> What about:
>
> file://foo,branch=foo/bar,a/file/in/the/branch
>
> What's wrong here?
That's a different path, it refers to the branch:

"branch" in directory "the" in directory "in" in directory "file" in
directory "bar,a" in the directory "foo" with branch parameters
"branch=foo".

Cheers,

Jelmer

[Alexander Belchenko (bialix)]

Jelmer Vernooij пишет:
> That's a different path, it refers to the branch:
>
> "branch" in directory "the" in directory "in" in directory "file" in
> directory "bar,a" in the directory "foo" with branch parameters
> "branch=foo".

I'm unable to parse it. Sorry.

[Neil Martinsen-Burrell (nmb)]

On Fri, Dec 23, 2011 at 10:04, Alexander Belchenko <email address hidden> wrote:
> Jelmer Vernooij пишет:
>> That's a different path, it refers to the branch:
>>
>> "branch" in directory "the" in directory "in" in directory "file" in
>> directory "bar,a" in the directory "foo" with branch parameters
>> "branch=foo".
>
> I'm unable to parse it. Sorry.

He means that the branch is named foo and the file path is
/bar,a/file/in/the/branch.

I can see that since we chose to use URL syntax for referring to
branch names, then we may be stuck with URL encoding for our branch
names which contain URL-reserved characters.

[Jelmer Vernooij (jelmer)]

The url syntax is available but it is by no means the only way to address a collocated branch. You can eg specify branch names directly to some commands already - e.g switch.
--
Sent from my phone. Please excuse my brevity.

Neil Martinsen-Burrell <email address hidden> wrote:

On Fri, Dec 23, 2011 at 10:04, Alexander Belchenko <email address hidden> wrote:
> Jelmer Vernooij пишет:
>> That's a different path, it refers to the branch:
>>
>> "branch" in directory "the" in directory "in" in directory "file" in
>> directory "bar,a" in the directory "foo" with branch parameters
>> "branch=foo".
>
> I'm unable to parse it. Sorry.

He means that the branch is named foo and the file path is
/bar,a/file/in/the/branch.

I can see that since we chose to use URL syntax for referring to
branch names, then we may be stuck with URL encoding for our branch
names which contain URL-reserved characters.

--
You received this bug notification because you are a member of bzr-core,
which is subscribed to Bazaar.
https://bugs.launchpad.net/bugs/907980

Title:
Should be able to create colocated branches with "/" in the name to
allow for hierarchical branch names

To manage notifications about this bug go to:
https://bugs.launchpad.net/bzr/+bug/907980/+subscriptions

[Jelmer Vernooij (jelmer)]

On 12/23/2011 06:25 PM, Neil Martinsen-Burrell wrote:
> On Fri, Dec 23, 2011 at 10:04, Alexander Belchenko <email address hidden> wrote:
>> Jelmer Vernooij пишет:
>>> That's a different path, it refers to the branch:
>>>
>>> "branch" in directory "the" in directory "in" in directory "file" in
>>> directory "bar,a" in the directory "foo" with branch parameters
>>> "branch=foo".
>>
>> I'm unable to parse it. Sorry.
>
> He means that the branch is named foo and the file path is
> /bar,a/file/in/the/branch.
>
> I can see that since we chose to use URL syntax for referring to
> branch names, then we may be stuck with URL encoding for our branch
> names which contain URL-reserved characters.
URLs aren't the only way you can reference colocated branches. For
example, "bzr switch" accepts 'plain' colocated branch names too. I
imagine more commands will support this sort of thing over time.

Cheers,

Jelmer

[Gordon Tyler (doxxx)]

In my opinion, while it would be nice and perhaps necessary for the internals to have branches addressable solely using a URL, it is not the most user friendly interface. I personally would much prefer to refer to a branch using a separate command-line option, or even have relative branch references try to resolve against colocated branches before going to look in the filesystem. e.g. `bzr merge my-feature` should look for a colocated branch called 'my-feature' before trying to find a sibling directory by that name.

So I guess what I'm saying is, don't worry too much about making the URL user-friendly because it never will be. Instead, effort should be put into making sure that all commands which can deal with branches have the ability to do so in a user-friendly manner either by relative reference or a command-line option.

[Jelmer Vernooij (jelmer)]

Am 24/12/11 16:05, schrieb Gordon Tyler:
> In my opinion, while it would be nice and perhaps necessary for the
> internals to have branches addressable solely using a URL, it is not the
> most user friendly interface. I personally would much prefer to refer to
> a branch using a separate command-line option, or even have relative
> branch references try to resolve against colocated branches before going
> to look in the filesystem. e.g. `bzr merge my-feature` should look for a
> colocated branch called 'my-feature' before trying to find a sibling
> directory by that name.
Right, that's exactly what the plan is. :-)

Cheers,

Jelmer

[Jelmer Vernooij (jelmer)]

Am 24/12/11 16:05, schrieb Gordon Tyler:
> In my opinion, while it would be nice and perhaps necessary for the
> internals to have branches addressable solely using a URL, it is not the
> most user friendly interface. I personally would much prefer to refer to
> a branch using a separate command-line option, or even have relative
> branch references try to resolve against colocated branches before going
> to look in the filesystem. e.g. `bzr merge my-feature` should look for a
> colocated branch called 'my-feature' before trying to find a sibling
> directory by that name.
Right, that's exactly what the plan is. :-)

Cheers,

Jelmer