bzr push / bzr merge

[Alexander Belchenko]

IIRC, in many cases push and pull transferring missing revisions and other data
first and only then check if branches are diverged. So, after this any operation
will be much faster. This explains why you see fast merge after push/pull:
all data is already in your repo.

I can be wrong, but at least I remember bzr worked this way in the past.

Parth Malwankar пишет:
> On Mon, Mar 30, 2009 at 3:07 PM, Parth Malwankar
> <parth.malwankar at gmail.com> wrote:
>> Hello,
>>
>> I use bzr+ssh to talk to my server which holds the repo.
>> I have a local branch which I keep in sync using bzr pull or merge
>> as needed. In case the server copy has diverged due to checkin
>> by someone else, push fails and then I do a merge.
>>
>> I recently upgraded to bzr 1.13 and observed that if I do a "bzr
>> push" on a branch that has diverged, bzr transfers about 2MB
>> of data before failing saying that I need to do a merge. It says
>> "walking tree 1/1" during this time.
>>
>> "bzr pull" itself is very quick and transfers about 50kB or so
>> based on the changeset size I think.
>>
> 
> Sorry, I meant "bzr merge" is very fast above.
> 
>> I was wondering if bzr pull transferring so much data before
>> failing is expected behaviour. Considering
>> that the branches have diverged and that bzr is just going
>> to tell me to merge instead of pull, I was confused as to why
>> so much data is transferred.
>>
>> In case the branches have not diverges, pull is very quick,
>> 50kB or so.
>>
>> I am using 1.9-rich-root on the server as well as client.
>> In case it matters, we migrated from svn a while back after
>> importing my repo and now use bzr exclusively for this
>> project.
>>
>> Thanks.
>> Parth
>>
> 
>