qannotate: crash when trying to annotate a deleted file.

[Marius Kruger]

This could be a resurrection of https://bugs.launchpad.net/qbzr/+bug/419790

I get this when trying to annotate a file that has been deleted, i.e. it is not in the working tree.
The commandline annotate can annotate the revision I'm interested in just fine.
As a workaround I branched the revision I'm interested in and qannotated that.

Steps to reproduce using the qbzr source from the above mentioned bug:

Open qlog in qbzr branch.
go to rev 240
Right click on "_ext.cpp => lib/_ext.cpp", click on annotate.

What happens: It crashes (using bzr and qbzr trunks):

Wed 2011-06-29 01:47:35 +0200
0.108 bazaar version: 2.4.0dev5
0.109 bzr arguments: [u'qlog']
0.148 looking for plugins in /home/amanica/.bazaar/plugins
0.361 looking for plugins in /stf_fast/prj/bzr/bzr.repo/bzr.dev/bzrlib/plugins
0.361 Plugin name launchpad already loaded
0.363 looking for plugins in /usr/lib/python2.7/dist-packages/bzrlib/plugins
0.363 Plugin name launchpad already loaded
0.363 Plugin name weave_fmt already loaded
0.363 Plugin name news_merge already loaded
0.363 Plugin name changelog_merge already loaded
0.363 Plugin name netrc_credential_store already loaded
0.363 Plugin name bash_completion already loaded
0.370 encoding stdout as sys.stdout encoding 'UTF-8'
0.716 opening working tree '/stf_fast/prj/bzr/plugins/qbzr'
12.398 Traceback (most recent call last):
  File "/home/amanica/.bazaar/plugins/qbzr/lib/util.py", line 908, in run_in_loading_queue
    f(*args, **kargs)
  File "/home/amanica/.bazaar/plugins/qbzr/lib/logmodel.py", line 116, in load
    super(WorkingTreeHasChangeFilter, self).load()
  File "/home/amanica/.bazaar/plugins/qbzr/lib/loggraphviz.py", line 1539, in load
    if self.has_changes(tree):
  File "/home/amanica/.bazaar/plugins/qbzr/lib/loggraphviz.py", line 1566, in has_changes
    for file_id in self.file_ids]
  File "/stf_fast/prj/bzr/bzr.repo/bzr.dev/bzrlib/decorators.py", line 154, in read_locked
    result = unbound(self, *args, **kwargs)
  File "/stf_fast/prj/bzr/bzr.repo/bzr.dev/bzrlib/workingtree_4.py", line 481, in id2path
    raise errors.NoSuchId(tree=self, file_id=file_id)
NoSuchId: The file id "_ext.cpp-20071223174633-d2tjggdxlg3hlbd7-1" is not present in the tree <WorkingTree6 of /stf_fast/prj/bzr/plugins/qbzr>.

12.398 Traceback (most recent call last):
  File "/home/amanica/.bazaar/plugins/qbzr/lib/util.py", line 908, in run_in_loading_queue
    f(*args, **kargs)
  File "/home/amanica/.bazaar/plugins/qbzr/lib/logmodel.py", line 116, in load
    super(WorkingTreeHasChangeFilter, self).load()
  File "/home/amanica/.bazaar/plugins/qbzr/lib/loggraphviz.py", line 1539, in load
    if self.has_changes(tree):
  File "/home/amanica/.bazaar/plugins/qbzr/lib/loggraphviz.py", line 1566, in has_changes
    for file_id in self.file_ids]
  File "/stf_fast/prj/bzr/bzr.repo/bzr.dev/bzrlib/decorators.py", line 154, in read_locked
    result = unbound(self, *args, **kwargs)
  File "/stf_fast/prj/bzr/bzr.repo/bzr.dev/bzrlib/workingtree_4.py", line 481, in id2path
    raise errors.NoSuchId(tree=self, file_id=file_id)
NoSuchId: The file id "_ext.cpp-20071223174633-d2tjggdxlg3hlbd7-1" is not present in the tree <WorkingTree6 of /stf_fast/prj/bzr/plugins/qbzr>.

[Alexander Belchenko (bialix)]

Based on traceback I see we're trying to get the file from workig tree. Apparetly deleted file is not here.

Could you say, please, have your tree had uncommitted changes?

[Marius Kruger (amanica)]

On 29 June 2011 09:59, Alexander Belchenko <email address hidden> wrote:

> Based on traceback I see we're trying to get the file from workig tree.
> Apparetly deleted file is not here.
>

Yes it has been deleted a number of revisions back.

> Could you say, please, have your tree had uncommitted changes?
>

Yes, I wanted to check something before committing, but I don't think that
makes a difference:
I could reproduce it now with the qbzr tree (with no uncommitted stuff) with
the steps in the original bug.

[Alexander Belchenko (bialix)]

Marius Kruger пишет:
> On 29 June 2011 09:59, Alexander Belchenko <email address hidden> wrote:
>> Could you say, please, have your tree had uncommitted changes?
>
> Yes, I wanted to check something before committing, but I don't think that
> makes a difference:
> I could reproduce it now with the qbzr tree (with no uncommitted stuff) with
> the steps in the original bug.

Thank you. I think it should be pretty easy to fix it: we just need to
use RevisionTree instead of WorkingTree somewhere. I just can't say off
hands where exactly.