Loggerhead fails to browse revisions that change binary files

[Andrew Bennetts]

On http://codebrowse.launchpad.net/~guillo.gonzo/+junk/org.vcs.bazaar.eclipse.old/changes, there's a link to http://codebrowse.launchpad.net/~guillo.gonzo/+junk/org.vcs.bazaar.eclipse.old/revision/guillermo%40steppenwolf.local-20070226051838-rotj3lixl1ydx1st?start_revid=guillermo%40steppenwolf.hiperborea-20070228013920-epjupqxw2nxhci5r (for revision 3). That page gives a 500 error.

[James Henstridge (jamesh)]

Here's the relevant traceback (with some of the path names trimmed for readability):

  File "CherryPy-2.2.1-py2.4.egg/cherrypy/_cphttptools.py", line 105, in _run
    self.main()
  File "CherryPy-2.2.1-py2.4.egg/cherrypy/_cphttptools.py", line 254, in main
    body = page_handler(*virtual_path, **self.params)
  File "loggerhead/util.py", line 344, in _f
    out = f(*a, **kw)
  File "<string>", line 3, in default
  File "TurboGears-1.0.1-py2.4.egg/turbogears/controllers.py", line 334, in expose
    output = database.run_with_transaction(
  File "<string>", line 5, in run_with_transaction
  File "TurboGears-1.0.1-py2.4.egg/turbogears/database.py", line 302, in so_rwt
    retval = func(*args, **kw)
  File "<string>", line 5, in _expose
  File "TurboGears-1.0.1-py2.4.egg/turbogears/controllers.py", line 351, in <lambda>
    mapping, fragment, args, kw)))
  File "TurboGears-1.0.1-py2.4.egg/turbogears/controllers.py", line 378, in _execute_func
    output = errorhandling.try_call(func, *args, **kw)
  File "TurboGears-1.0.1-py2.4.egg/turbogears/errorhandling.py", line 73, in try_call
    return func(self, *args, **kw)
  File "loggerhead/controllers/revision_ui.py", line 73, in default
    change = h.get_changes([ revid ], get_diffs=True)[0]
  File "loggerhead/util.py", line 334, in locked
    return unbound(self, *args, **kw)
  File "loggerhead/history.py", line 596, in get_changes
    changes = self._change_cache.get_changes(revid_list, get_diffs)
  File "loggerhead/util.py", line 334, in locked
    return unbound(self, *args, **kw)
  File "loggerhead/changecache.py", line 107, in get_changes
    changes = self.history.get_changes_uncached(fetch_list, get_diffs)
  File "loggerhead/util.py", line 334, in locked
    return unbound(self, *args, **kw)
  File "loggerhead/history.py", line 67, in bzrlib_read_locked
    return unbound(self, *args, **kw)
  File "loggerhead/history.py", line 693, in get_changes_uncached
    entry.changes = self.parse_delta(delta, get_diffs, old_tree, new_tree)
  File "loggerhead/util.py", line 334, in locked
    return unbound(self, *args, **kw)
  File "loggerhead/history.py", line 809, in parse_delta
    handle_modify(path, path, fid, kind)
  File "loggerhead/history.py", line 801, in handle_modify
    bzrlib.diff.internal_diff(old_path, old_lines, new_path, new_lines, buffer)
  File "bzrlib/diff.py", line 70, in internal_diff
    textfile.check_text_lines(oldlines)
  File "bzrlib/textfile.py", line 42, in check_text_lines
    raise BinaryFile()
BinaryFile: File is binary but should be text.

Looks like Loggerhead does not handle changes to binary files well.

[James Henstridge (jamesh)]

Open a loggerhead task too, since this will need fixing there too.

[James Henstridge (jamesh)]

Robey: the linked branch contains the fix I'm rolling out for codebrowse.launchpad.net. It should be appropriate to apply to your loggerhead tree too.

[James Henstridge (jamesh)]

Fixed for codebrowse.launchpad.net

[Robey Pointer (robey)]

merged!

[Michael Hudson-Doyle (mwhudson)]

Released in 1.2.