Re: [cti-stix] RE: STIX 2.0 Specification Questions

[John-Mark Gurney <jmg@newcontext.com>]

Jason Keirstead wrote this message on Fri, Aug 12, 2016 at 11:03 -0300:
> For the record - I agree with Allan on this. We should not be attempting to
> define when someone should version an object. This is up to the tool
> implementer / producer of the intel and is highly context-specific (we will
> never get it right).

I agree that we cannot define when somone should version an object.

I'm just saying that we shouldn't break versioning because someone's
new "version" of an object conflicts w/ a relationship.  Either they
should have created a different object, OR they could have revoked
the relationship before/at same time, as the new object...

What happens when someone revisions COA or something, and then deploys
an Indicator that references it?  If we allow a free for all on
which version of the object is used, that really creates confusion
on the producer for what the consumer will use/see.  This could actually
hamper people from using versioning.

Oh, they are using a product that is known not to work w/ versioning,
so instead of versioning an object, we'll just always create new
objects, etc.

Maybe this more should be in an implementors guide, but as we don't
have a document where we can put this information, where else are
we going to put it?

-- 
John-Mark