[PATCH 0 of 6] introduce peer interface

[Peter Arrenbrecht]

On Wed, Jun 8, 2011 at 6:49 PM, Peter Arrenbrecht
<peter.arrenbrecht at gmail.com> wrote:
> On Wed, Jun 8, 2011 at 6:45 PM, Peter Arrenbrecht
> <peter.arrenbrecht at gmail.com> wrote:
>> This patch queue introduces the "peer" interface for views of possibly remote repositories.
>> This opens the way to differentiating the behaviour of local and remote calls, and also allows
>> us to start keeping their namespaces apart. For instance, I have already factored the legacy
>> method out of the normal localpeer. So a client talking to local repos is now forced to use only
>> modern peer API calls.
>>
>> We could now start to use different names for local and remote methods to ensure that we
>> properly use peers and repos even in tests that run on local repos (duck typing is a pain
>> here). We could also move a couple of methods to the peer outright (like known(), for
>> instance).
>>
>> Finally we can now consider different semantics for local and peer calls, for example in
>> getbundle(). getbundle() on a localrepo will mostly be fed into some sort of stream. getbundle()
>> on a peer will usually be passed to addchangegroup, or else written to a bundle file.
>
> If this makes it in, I will then add the batching support to peer.py,
> which is finally the right place for it, I hope.

Matt, by the way, feel free to meddle with these patches as you see
fit. I'd just like somewhere to finally get the batching code in so we
can get both subset directions done in one roundtrip.
-parren