parallel_emerge: Fix builds when PORTAGE_BINHOST is unset

parallel_emerge: Fix builds when PORTAGE_BINHOST is unset

BUG=none
TEST=Try building for a branch new board overlay that doesn't already have a binhost setup.

[David James, 2010-07-26 10:23:14]

Can we just catch the case where PORTAGE_BINHOST is undefined, and succeed on
that case, but throw errors on other cases?

[Olof Johansson, 2010-07-26 15:38:50]

On 2010/07/26 10:23:14, David James wrote:
> Can we just catch the case where PORTAGE_BINHOST is undefined, and succeed on
> that case, but throw errors on other cases?

Hmm. Do we care about cases where the build machine is offline and the url fetch
would fail because of that?

[Olof Johansson, 2010-07-27 16:39:17]

David suggested I only catch the ValueError exception to still err out on other
problems. But I started looking at the code, and it doesn't make sense to call
RemotePackagDatabase() if PORTAGE_BINHOST is unset, so check that before calling
instead.

PTAL.

[davidjames, 2010-07-27 16:44:03]

LGTM w/nits

http://codereview.chromium.org/2870068/diff/5001/6001
File parallel_emerge (right):

http://codereview.chromium.org/2870068/diff/5001/6001#newcode956
parallel_emerge:956: if self.emerge.settings["PORTAGE_BINHOST"]:
This guy can just use emerge.settings since emerge is a local variable as well.
See above.

http://codereview.chromium.org/2870068/diff/5001/6001#newcode957
parallel_emerge:957: remote_pkgs = RemotePackageDatabase()
Maybe we can update RemotePackageDatabbase to accept the portage_binhost url as
an arg? That way, there's no need to reference settings["PORTAGE_BINHOST"]
twice.

http://codereview.chromium.org/2870068/diff/5001/6001#newcode958
parallel_emerge:958: else:
Please kill the tabs

[Olof Johansson, 2010-07-27 16:59:21]

Ok, based on that it seems a little cleaner to just move the test of empty
PORTAGE_BINHOST down into RemotePackageDatabase.

PT(yet)AL.

[davidjames, 2010-07-28 16:46:07]

LGTM