TrustEngine refactoring

[Cantor, Scott]

On 8/21/14, 7:12 AM, "Rod Widdowson" <rdw at steadingsoftware.com> wrote:

>> I'm not very fluent in the Spring parsing code. I guess what we probably
>> should do is actually parse the existing trust engine(s) (which I guess
>> we're doing for metadata anyway?)
>
>We only parse those trust engines which Metadata parsing needs.  If we
>meet
>any of the following we issue log.warn.

Yes, I see. The xsi:type names are quite confusing since the MetadataXXX
engines are the ones that use metadata but are not used for/by metadata.

>>But I don't know how much we care. We could hardwire the two chaining
>> engines into the system when we parse legacy and just emit warnings like
>> with the old security policy stuff.
>
>I don't feel qualified to answer which is the correct thing to do - I'll
>defer to want you and Brent decide.

I think it ends up being pointless to parse them, because the place they
get used in the legacy schema is to reference them from policy rules,
which we're not parsing. In other words, if somebody had customized the
default engines, we couldn't properly inject the result of parsing them
into the right places to make it work without following a ton of
references.

Given the very small chance anybody would ever have done this, and the
fact that we can't really replicate what they did, I think it's better to
warn and document, and just install a default set using
profile-defaults.xml

-- Scott