Robert Collins wrote:
> Review: Approve
> review: +1
>
> The use of vf.keys() in the gc pack code - might want to check that
> reconcile() doesn't use revision selectors - as in that case you will
> copy too many inventories [if all tests are passing this probably isn't
> the case]
>
> -Rob
>
Reconcile has its own implementation of _copy_inventory_texts:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Robert Collins wrote:
> Review: Approve
> review: +1
>
> The use of vf.keys() in the gc pack code - might want to check that
> reconcile() doesn't use revision selectors - as in that case you will
> copy too many inventories [if all tests are passing this probably isn't
> the case]
>
> -Rob
>
Reconcile has its own implementation of _copy_inventory _texts:
def _copy_inventory _texts( self): vfs('inventory' , True, True)
self._ copy_stream( source_ vf, target_vf, self.revision_keys,
'inventori es', self._get_ filtered_ inv_stream, 2)
self. _data_changed = True
source_vf, target_vf = self._build_
if source_vf.keys() != self.revision_keys:
Which is saying "I'm going to prune down to the minimum, and if the
source's number of keys is different than the minimum, then things have
changed."
This means that reconcile on stacked branches is probably slightly wrong.
But anyway, this change obviously doesn't effect reconcile :).
John enigmail. mozdev. org/
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAkq DKfUACgkQJdeBCY SNAAOuKQCeKD4Pt Kpnw6Ro0AZaWv/ w0LPp 3kDA39hojP67PXD b715bJXIKU
6woAoI/
=T66k
-----END PGP SIGNATURE-----