[RFC] Ways to make initial knit creation faster (both push and commit)

[Robey Pointer]

On 11 Sep 2006, at 8:40, John Arbash Meinel wrote:

> Robey Pointer wrote:
>>
>
> ...
>
>>> We recently upgraded bzr.dev to run the tests with a separate  
>>> HOME and
>>> work directories @1991. Can you update and see if all the tests  
>>> pass? We
>>> already removed the 02666 test, since it was bogus.
>>>
>>> All the tests now pass on my Mac. Can you check for you?
>>
>> There's one new failure (something expected a different sftp log  
>> line)
>> but the other 6 are gone, so overall a good improvement. :)
>>
>> AssertionError: '' != 'Opened sftp connection (server version 3)\n'
>>
>> robey
>
> Well, that sounds like paramiko changed how it was logging  
> information.
> Not a bzr change, but a paramiko one. Can you tell me what test is  
> failing?

Never mind, that turned out to be from a modification I'd made to  
bzr.dev for debugging.  I keep forgetting that just because "bzr  
pull" succeeds, that doesn't mean you don't have possible uncommitted  
changes.

Looks like all the unit tests pass now! :)

robey