bzr.dev missing signatures

[John A Meinel]

John A Meinel wrote:
> Erik Bågfors wrote:
>> On 5/8/06, John A Meinel <john at arbash-meinel.com> wrote:
>>> Erik Bågfors wrote:
>>> > After reading this, I thought I should sign my commits. It felt 
>>> natural
>>> > to run
>>> > sign-my-commits.  It dies with the following
>>> >
>>> > : [bagfors at zyrgelkwyt]$ ; bzr sign-my-commits
>>> > erik at bagfors.nu-20051220112436-0c74ef0e272fe074
>>> > bzr: ERROR: exceptions.UnicodeDecodeError: 'ascii' codec can't decode
>>> > byte 0xc3 in position 10: ordinal not in range(128)
>>> >  at /home/bagfors/src/bzr/bzr.dev/bzrlib/testament.py line 139
>>> >  in as_text_lines
>>> >
>>> > /Erik
>>>
>>> What version of bzr are you using? Because I know there was a patch
>>> around that fixed testament to properly encode things as utf-8, and I'm
>>> hoping we don't have a regression.
>>
>> bzr.dev latest (which is 0.8 :) )
>>
>>
>>> I'll look through my encoding branch, and see if testament fails with
>>> Unicode committers.
>>
>> Ok
>>
> 
> Can you send me a more complete backtrace?
> 
> The line in question is doing:
> return [line.encode('utf-8') for line in r]
> 
> So it is supposed to be encoding things properly. It would seem that 
> somewhere internally we are actually pre-encoding into utf-8, and then 
> we are encoding again into utf-8 (which tries to decode as ascii first, 
> and fails).

Just to confirm my suspicion, at the very least the call:
_revprops_to_lines() does encode into utf-8, So if there was a unicode 
entry in revision properties, that would cause a double utf-8 encode.

I'm not sure whether to fix things in my encoding branch (which now has 
a chance of landing, but isn't guaranteed for a while), or whether to do 
it on an independent branch (which has a much higher chance of actually 
landing soon).

John
=:->