[MERGE][bug #300177] Pass compression_parent as part of 'get_record_stream'

[Robert Collins]

On Mon, 2009-02-16 at 11:13 -0600, John Arbash Meinel wrote:
> 
> Well, the base level ContentFactory implements "compression_parents =
> None", so at least implementations that inherit from that wouldn't
> have
> to change.
> 
> I still haven't heard a great way to actually solve this. If I make it
> an API on ContentFactory then the same argument applies. It is either
> that or change "insert_record_stream" to do "getattr()" logic, which
> seems a bit overkill.

insert_record_stream already has type safe dispatch via the storage_kind
and the adapter registry, there is no need to add anything new.

> In essence, we need *some* way to determine if a record refers to
> another record before it can be properly extracted to a fulltext. We
> don't currently have that (well we do, but infer it from a list of
> known
> types, which is a *worse* solution IMO).

Actually that is the intent of the design; and I'm using that heavily in
doing network serialisation at the moment - have a look at my recent
commits on the commit list. Just addressing the delta_closure=True case
over the network and my branch will be going up for review.

Having got back into this code, I'm *really* against putting
compression_parent as a top level component of the API. Its irrelevant
for more cases than it is relevant.

-Rob
-------------- 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/20090217/57c2fc66/attachment.pgp