FWIW, the error John reported on the mailing list is:
ErrorFromSmartServer: Error received from smart server: ('error', "bytes
must be a string, got <type 'bzrlib._static_tuple_c.StaticTuple'>")
So something is passing a StaticTuple to _bytes_to_text_key, which
sounds like some code already has a key but thinks it has the bytes
representation it. The callers of that are in
bzrlib/repofmt/groupcompress_repo.py; the only one of those I'd expect
to be triggered on stream insertion is in
GCRepositoryPackCollection._check_new_inventories, which is invoked at
the end of inserting a stream, which matches the original report here
and John's report on the mailing list.
The mystery then is why that code path is failing for them, because I'm
pretty sure basically every write to a 2a repository would exercise that
call, so presumably something earlier during receiving the stream is
getting muddled. There's nothing obviously wrong with that code that I
can see, and it's especially weird that it's only happening for some
people sometimes, rather than constantly.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/721163
Title:
bzr crashed with ErrorFromSmartServer: ('error', 'bytes must be a
string')
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
