Comment 27 for bug 98836

These locks are not enough to keep the local bazaar repository in shape any way!

******************************************************************************
$ bzr ci src/ksn/Makefile
Committing to: <repo_path>
aborting commit write group: DirstateCorrupt(The dirstate file (DirState(u'<repo_path>/.bzr/checkout/dirstate')) appears to be corrupt: Bad parse, we expected to end on \n, not: 45 to_and_nct-20120307150343-sa5f7rw4ganyg7xx-21: (StaticTuple('tests/etc', 'Spec_Plus_Nums', 'spec_plus_nums-20120307150343-sa5f7rw4ganyg7xx-20'), [StaticTuple('l', '<other_path>/forPYuxwAAAAJAD+DXAAAKH/', 0L, 0, '0'), StaticTuple('n', '<email address hidden>', 0L, 0, 'TO_And_NCT')]))
bzr: ERROR: The dirstate file (DirState(u'<repo_path>r/.bzr/checkout/dirstate')) appears to be corrupt: Bad parse, we expected to end on \n, not: 45 to_and_nct-20120307150343-sa5f7rw4ganyg7xx-21: (StaticTuple('tests/etc', 'Spec_Plus_Nums', 'spec_plus_nums-20120307150343-sa5f7rw4ganyg7xx-20'), [StaticTuple('l', '<other_path>/forPYuxwAAAAJAD+DXAAAKH/', 0L, 0, '0'), StaticTuple('n', '<email address hidden>', 0L, 0, 'TO_And_NCT')])
******************************************************************************

Yes, we have some kind of problem in our NFS setup.
Yes, I interrupted a "status" command right before the above happend.

No, I _do_not_ expect a _read_ command to be able to trash my repository.

Could you possibly stop updating files in place and write the new information to a backup file and move it to the right name after it is complete?
(That's what I/we always do to avoid these kind of problems. Move or better yet hard-links use to be atomic in the file-system.)