Code review comment for lp:~vila/bzr/config-lock-remote

>>>>> John Arbash Meinel writes:

<snip/>

    > Getting at AttributeError is an ugly way to say the branch has
    > disappeared. Very hard to discover when it happens in production
    > (accidentally certainly, but still possible).

    > So please consider giving us a nicer error for discoverability purposes.

Could you point me where in the *actual* code this idiom is used and
such a reporting is taken into account ?

I find it highly unproductive to be blocked from landing incremental
patches when my intent is to ease review.

This code is *not* used in production right now and when it will this
case will never be triggered (and could not possibly be). The only cases
now and tomorrow are unit tests which purposefully *avoid* creating such
objects *without* a protecting branch. And these tests already have a
comment explaining the issue.