Ian Clatworthy wrote:
> Review: Needs Information
> Thanks a bucket-load for these changes. I'll take a detailed look next Monday when my brain is a little more awake.
>
> FWIW, my main concern is bzr version compatibility. Until now, fastimport has supported early versions of bzr: 1.1 according to README.txt. I'm happy to bump that to 2.0.0 (say) but I don't want to depend on any APIs only in the 2.1.0 branch yet. Off the top of your head, do we need to tweak any of this patch accordingly?
I don't know of anything, but it is best to run an import with an older
bzr to make sure :).
There is at least one bugfix that landed in trunk, but the code already
checks for that
if isinstance(builder.inv_sha1, tuple):
...
KnownGraph was introduced in bzr 1.17, and though 'add_node' is new, it
also has a 'getattr()' check around that (though I think that is a
different patch come to think of it.)
I think that if the switch to CommitBuilder is finalized, then we should
look into cleaning up the code, as you have a lot of duplicate logic in
there.
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Ian Clatworthy wrote:
> Review: Needs Information
> Thanks a bucket-load for these changes. I'll take a detailed look next Monday when my brain is a little more awake.
>
> FWIW, my main concern is bzr version compatibility. Until now, fastimport has supported early versions of bzr: 1.1 according to README.txt. I'm happy to bump that to 2.0.0 (say) but I don't want to depend on any APIs only in the 2.1.0 branch yet. Off the top of your head, do we need to tweak any of this patch accordingly?
I don't know of anything, but it is best to run an import with an older
bzr to make sure :).
There is at least one bugfix that landed in trunk, but the code already
checks for that
if isinstance( builder. inv_sha1, tuple):
...
KnownGraph was introduced in bzr 1.17, and though 'add_node' is new, it
also has a 'getattr()' check around that (though I think that is a
different patch come to think of it.)
I think that if the switch to CommitBuilder is finalized, then we should
look into cleaning up the code, as you have a lot of duplicate logic in
there.
John
=:->
-----BEGIN PGP SIGNATURE----- enigmail. mozdev. org/
ZIrEACgkQJdeBCY SNAANViwCg1d18S 7ygPWdG0e89xvSa 0IXK 8+PxuMIHShsPpuO P7
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAks
UjcAoIKWth4vZzo
=cwJW
-----END PGP SIGNATURE-----