On 11/04/2011 05:44 PM, Eli Zaretskii wrote:
> Yes, it's reproducible. The last revno I have is 10926. This branch
> has no changes at all, not even changes made with bzr. It's a pristine
> mirror of the upstream. All I ever did in it is "bzr pull" and commands
> like "bzr log" that are read-only.
>
> I copied the branch aside, downgraded to bzr-git 0.6.2, and then "bzr
> pull" worked without any problems (and without churning the disk).
> (However, it is slow and inefficient: it fetches 30+MB of data just to
> import a small number revisions.) So it sounds like some change in
> later bzr-git, perhaps related to the HTTP-related improvements made
> lately.
>
> Let me know if I can provide more details. Perhaps some debugging
> switch will help?
Can you reproduce this issue with 0.6.3?
It sounds like this might be related to the smart server protocol
support that was introduced in 0.6.3. That would also explain why it's
so inefficient for your with 0.6.2.
On 11/04/2011 05:44 PM, Eli Zaretskii wrote:
> Yes, it's reproducible. The last revno I have is 10926. This branch
> has no changes at all, not even changes made with bzr. It's a pristine
> mirror of the upstream. All I ever did in it is "bzr pull" and commands
> like "bzr log" that are read-only.
>
> I copied the branch aside, downgraded to bzr-git 0.6.2, and then "bzr
> pull" worked without any problems (and without churning the disk).
> (However, it is slow and inefficient: it fetches 30+MB of data just to
> import a small number revisions.) So it sounds like some change in
> later bzr-git, perhaps related to the HTTP-related improvements made
> lately.
>
> Let me know if I can provide more details. Perhaps some debugging
> switch will help?
Can you reproduce this issue with 0.6.3?
It sounds like this might be related to the smart server protocol
support that was introduced in 0.6.3. That would also explain why it's
so inefficient for your with 0.6.2.
Cheers,
Jelmer