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

David Binger wrote:
> But it can be avoided.  We traverse the object records, writing them out
> as we find them to the new packed file.  The index of the new packed
> file is currently kept in RAM, but we can avoid that by constructing
> and writing the new index during a direct scan of the packed records.
> I don't see any obstacle to doing this with a constant amount of RAM.

You do need a way to avoid multiple copies of an already copied object.
Yes, you can use a file on disk for keep that info around.

Still, if I do a garbage collection of a 5 Gigabytes filestorage, I will
need a temporal 5 gigabytes copyspace on disk. And remember, my storage
is about 480 gigabytes, currently. No way I could do a garbage
collection in the way currently durus backend works :-p : performance,
time use, disk bandwidth and I can't afford a 480 mbytes temporal file...

So the reference counters in the BerkeleyDB backend :-)

- --
Jesus Cea Avion                         _/_/      _/_/_/        _/_/_/
jcea@argo.es http://www.argo.es/~jcea/ _/_/    _/_/  _/_/    _/_/  _/_/
jabber / xmpp:jcea@jabber.org         _/_/    _/_/          _/_/_/_/_/
                               _/_/  _/_/    _/_/          _/_/  _/_/
"Things are not so easy"      _/_/  _/_/    _/_/  _/_/    _/_/  _/_/
"My name is Dump, Core Dump"   _/_/_/        _/_/_/      _/_/  _/_/
"El amor es poner tu felicidad en la felicidad de otro" - Leibniz
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRCAsp5lgi5GaxT1NAQK+TAP/THLoSn5SM8ABsrqtTaAz+5TWMBCjTMBC
y/qCeJ1dUcQ8Tpjf5UkvxwDq1MxKECHg2z0oXMmAgv/6VMsRgLUR8MzEc+Lb61Ey
ATM+QPDy7RlD+J4PlTeiQu76Cztu10IUnwwnHLKzmk6ygMjDu8ZYRofK3byVjLdu
yBS8Xjwrn0Y=
=0tOS
-----END PGP SIGNATURE-----