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

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Collins wrote:
> Robert Collins has voted comment.
> Status is now: Semi-approved
> Comment:
> I'd be a lot happier with this if the compression_parent field was only
> on KnitContentFactory, which is the only content factory for which it
> makes sense.

I would have thought that having "compression_parent = None" would be
sufficient.

The issue is that insert_record_stream() already assumes it needs to
know about a compression parent (if type in deltas).

The main changes are:

use record.compression_parents to determine if this is a delta record,
rather than building up a delta types set.

Allow "compression_parents" to be a list rather than a single entry, for
potential things like mpdiffs, etc.

Personally, I think doing:

if record.compression_parents:
  # a delta

is better than

delta_types = set(['x', 'y', 'z'])
if record.storage_kind in delta_types:
  # a delta

Because the latter needs more maintenance than the former.

I'm happy to change it as long as we have a nice way to check that
something *is* a delta, such that we want to make sure to unpack it to a
fulltext if we are in a stacked repository.

I could add a new "ContentFactory.is_safe_in_stacked()" or something
like that.... doesn't sound better than what I've done, though.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkmQlAoACgkQJdeBCYSNAAMLOQCfSL6W5OpsilpdNyuj1df8odxD
J0wAn31iBCt7mxyRfKvY/epaybgypN60
=hr6J
-----END PGP SIGNATURE-----