[MERGE] Create a smart verb for Repository.set_make_working_trees.

Robert Collins robert.collins at canonical.com
Thu Feb 19 21:59:59 GMT 2009

On Thu, 2009-02-19 at 08:50 -0600, John Arbash Meinel wrote:
> Hash: SHA1
> Robert Collins wrote:
> > This removes a cause of probing for concrete repositories over the wire
> > during push of new branches and init of repositories.
> > 
> > It drops the count of rpc calls for init-repo to 16, and pushing a
> > trivial branch to 74.
> > 
> > -Rob
> > 
> Could we instead change the "initialize()" calls to set this? That seems
> like a lot better use of our time, since it means we have 1 round trip
> instead of 2 in the "optimal" case.
> If you feel that is too difficult, this patch seems fine.

I'll land it as is (but here is why):
 - bzr reconfigure does support changing this in the future anyhow, so 
   this RPC helps bzr reconfigure too.
 - its a bigger change to get init-repo to pass this into the stack
   earlier rather than calling it later, and will only save 1 round trip
   out of 5 (the blackbox layer takes 5 calls minimum).
 - due to invoking info, the actual case of creating a new repo would
   still be 15 calls even with doing that.

So I think that its not worth the refactoring (and slight loss of
clarity) to just push this into init right now. We're still halving the
number of RPC calls made by landing this and the predecessor patches,
which is a hopefully noticable win - but the really big wins are the
stream results on Branch.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20090220/94e067a0/attachment.pgp 

More information about the bazaar mailing list