On 2012/06/11 08:22:47, fwereade wrote:
> I still doubt this will have great value in practice, but LGTM.
Sorry Frank, I have to agree with William, even though the suggestion to
prevent this from happening was mine in the first place.
My misguided suggestion was based on the fact I had wrong memories
regarding the way we handle "deploy serviceA:foo serviceB:bar" commands.
We don't create the relation endpoint directly, but rather call a method
to resolve what we call the *descriptor* "<service>[:<relname>]". This
is the place that must have that validation to prevent errors from the
user from leaking through.
On 2012/06/11 08:22:47, fwereade wrote:
> I still doubt this will have great value in practice, but LGTM.
Sorry Frank, I have to agree with William, even though the suggestion to
prevent this from happening was mine in the first place.
My misguided suggestion was based on the fact I had wrong memories [:<relname> ]". This
regarding the way we handle "deploy serviceA:foo serviceB:bar" commands.
We don't create the relation endpoint directly, but rather call a method
to resolve what we call the *descriptor* "<service>
is the place that must have that validation to prevent errors from the
user from leaking through.
I apologize for wasting your time on this.
https:/ /codereview. appspot. com/6305067/