commit --exclude should NOT exclude TREE_ROOT

[Alexander Belchenko]

If branch has no commits, then --exclude `bzr root` fails. It's working OK if there is at least one commit.

C:\Temp>mkdir 4

C:\Temp>cd 4
C:\Temp\4>

C:\Temp\4>bzr init
Standalone tree (format: pack-0.92)
Location:
  branch root: .

C:\Temp\4>bzr ci -m test --exclude . --unchanged
Committing to: C:/Temp/4/
bzr: ERROR: exceptions.AssertionError: Root entry should be supplied to record_entry_contents, as of bzr 0.10.

Traceback (most recent call last):
  File "bzrlib\commands.pyo", line 893, in run_bzr_catch_errors
  File "bzrlib\commands.pyo", line 839, in run_bzr
  File "bzrlib\commands.pyo", line 539, in run_argv_aliases
  File "bzrlib\builtins.pyo", line 2431, in run
  File "bzrlib\decorators.pyo", line 192, in write_locked
  File "bzrlib\workingtree_4.pyo", line 237, in commit
  File "bzrlib\decorators.pyo", line 192, in write_locked
  File "bzrlib\mutabletree.pyo", line 201, in commit
  File "bzrlib\commit.pyo", line 377, in commit
  File "bzrlib\repository.pyo", line 159, in finish_inventory
AssertionError: Root entry should be supplied to record_entry_contents, as of bzr 0.10.

bzr 1.9 on python 2.5.2 (win32)
arguments: ['C:\\Program Files\\Bazaar\\bzr.EXE', 'ci', '-m', 'test', '--exclude', '.', '--unchanged']
encoding: 'cp1251', fsenc: 'mbcs', lang: 'ru'
plugins:
  acad C:\work\Bazaar\plugins\acad [0.6.1]
  bzrtools C:\Program Files\Bazaar\plugins\bzrtools [1.9.1]
  launchpad C:\Program Files\Bazaar\plugins\launchpad [unknown]
  qbzr C:\work\Bazaar\plugins\qbzr [0.9.6dev0]
  rebase C:\work\Bazaar\plugins\rebase [0.4.1]
  repo_push C:\work\Bazaar\plugins\repo_push [1.1.1]
  svn C:\Program Files\Bazaar\plugins\svn [0.4.14]
  x_bit C:\work\Bazaar\plugins\x_bit [unknown]
*** Bazaar has encountered an internal error.
    Please report a bug at https://bugs.launchpad.net/bzr/+filebug
    including this traceback, and a description of what you
    were doing when the error occurred.

[Jelmer Vernooij (jelmer)]

Interesting bug..

  status triaged
  importance medium
--
Jelmer Vernooij <email address hidden> - http://samba.org/~jelmer/
Jabber: <email address hidden>

[Robert Collins (lifeless)]

> C:\Temp\4>bzr ci -m test --exclude . --unchanged
> Committing to: C:/Temp/4/
> bzr: ERROR: exceptions.AssertionError: Root entry should be supplied to record_entry_contents, as of bzr 0.10.

Fun!

So, the base problem is that the first commit has to add the root of the
tree - we don't support trees with /no/ root.

And the exclude code is permitted to exclude everything. So, I consider
this a misleading error message, but not a bug - that commit *is not
valid*.

 status confirmed
 importance medium

[Alexander Belchenko (bialix)]

How's nice! It's not bug -- it's feature.

It's a leaking abstraction. Leaking in the face of user. The same command-line for any other commit works fine. Only first commit is so special. You need to commit TREE_ROOT always? Why you doing it implicitly on first user commit? Why not doing it explicitly on branch initialization? Automatic explicit commit by bzr itself to froze TREE_ROOT.

:-/

"Special cases aren't special enough to break the rules" ?

[Martin Pool (mbp)]

2009/11/4 Alexander Belchenko <email address hidden>:
> How's nice! It's not bug -- it's feature.
>
> It's a leaking abstraction. Leaking in the face of user. The same
> command-line for any other commit works fine. Only first commit is so
> special. You need to commit TREE_ROOT always? Why you doing it
> implicitly on first user commit? Why not doing it explicitly on branch
> initialization? Automatic explicit commit by bzr itself to froze
> TREE_ROOT.

So you think maybe this command should succeed and silently add the
tree root, so to the user it looks like a command that changed no
files?

--
Martin <http://launchpad.net/~mbp/>

[Alexander Belchenko (bialix)]

Martin Pool пишет:
> 2009/11/4 Alexander Belchenko <email address hidden>:
>> How's nice! It's not bug -- it's feature.
>>
>> It's a leaking abstraction. Leaking in the face of user. The same
>> command-line for any other commit works fine. Only first commit is so
>> special. You need to commit TREE_ROOT always? Why you doing it
>> implicitly on first user commit? Why not doing it explicitly on branch
>> initialization? Automatic explicit commit by bzr itself to froze
>> TREE_ROOT.
>
> So you think maybe this command should succeed and silently add the
> tree root, so to the user it looks like a command that changed no
> files?

Yes, it was my natural expectation. I have no way to control the TREE_ROOT entry from command-line,
as a plain user I even should not know that TREE_ROOT exists, right? In this case I'd expect to see
all machinery with TREE_ROOT working behind the scene.

If I'm wrong, please explain why and where.

[Martin Pool (mbp)]

To me at the cli level that would be reasonable.