Efficient revision fetching combined with chunked bodies

[Andrew Bennetts]

This diff isn't ready for merge yet, but I think it might be of interest.

It's basically the combination of two branches I've submitted already: the smart
method for efficient pulling of a set of revisions, plus the chunked body
streaming for the smart protocol.

Here's the diff of changes on top of those two branches.  It does some nasty
things to workaround a deficiency in the ContainerReader interface, but it seems
to work quite well.  I've been dogfooding it without any problems.

This ought to provide reasonable resource consumption (on the server and client)
for transferring a large amount of revision data, even for the initial pull of a
large branch, while still only having a small number of network roundtrips.

The branch can be found at:

    http://people.ubuntu.com/~andrew/bzr/streaming-smart-fetch

-Andrew.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: streaming-smart-fetch.diff
Type: text/x-diff
Size: 8961 bytes
Desc: not available
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070904/7cf456ec/attachment.bin