qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alan Conway" <acon...@redhat.com>
Subject Re: Review Request 39596: DISPATCH-186 - Add an indication that an annotation can be referenced by name
Date Fri, 30 Oct 2015 13:39:14 GMT


> On Oct. 30, 2015, 12:02 p.m., Justin Ross wrote:
> > I don't understand the true constraints here.  It seems to be a UI-driven one, about
what should be listed in the UI and what should not be, so I'd search more in the direction
of indexed=false or listed=false or hidden=true.
> > 
> > In any case, I think "referential" is the wrong word.  It's (1) obscure and (2)
doesn't unambiguously indicate the referent in this usage.  I guess referent=true would, but
again, too weird.  If you go with some variant of refer, I'd go referenceable.  A third grader
would understand that.

Background: we have managed "entities" which are things with attributes, such as connections
or connectors. Some of those entities (in particular connections/connectors) have fairly complicated
"groups" of related attributes that often have the same settings for multiple entities, for
examle the same SSL settings for many connections. We call those groups "annotations" after
the AMQP management spec (another poorly chosen word) and in a config file you can specify
the group of attributes once and name it, then refer to it by name in several entities as
a shorthand compared to repeating all the attribute values. The original UI just shows each
entity with all its attributes flattened out, but it is easier to read e.g. the SSL attributes
by turning them into a named group because the attribute values themselves are long and usually
not that interesting, it is more interesting to quickly identify which connections have the
*same* set of attributes. However not all things that are "annot
 ations" in the schema are interesting in this way, hence the search for a way to identify
the ones that are.

Its very possible that some rework of the schema could help too, to align the use of annotations
more closely with "things that are interesting to name as groups" so they could be one and
the same. Annotations evolved by munghing ideas in the original dispatch config file with
related ideas in the AMQP management spec, they may not have reached their ideal form yet.


- Alan


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/39596/#review104553
-----------------------------------------------------------


On Oct. 27, 2015, 4:31 p.m., Ernie Allen wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/39596/
> -----------------------------------------------------------
> 
> (Updated Oct. 27, 2015, 4:31 p.m.)
> 
> 
> Review request for qpid, Alan Conway, Ganesh M, Kenneth Giusti, mick goulish, and Ted
Ross.
> 
> 
> Repository: qpid-dispatch
> 
> 
> Description
> -------
> 
> Adds a new attribute to entities named referential. If true then the entity/annotation
could be referred to by name. This is to give the console enough information to separate out
the sslProfile attributes.
> 
> schema.py can already handle the case where a listener/connector contains a ssl-profile=<sslProfileName>
attribute.
> 
> I chose the name 'referential' to indicate that an annotation can be referred to by name.
Another possibility is 'referable'.
> 
> I also added an "references" list to an entity in the JSON schema. This list is only
emitted if any of the entity's annotations are marked as referential.
> 
> 
> Diffs
> -----
> 
>   python/qpid_dispatch/management/qdrouter.json c5b1edb 
>   python/qpid_dispatch_internal/management/schema.py 8f7e961 
> 
> Diff: https://reviews.apache.org/r/39596/diff/
> 
> 
> Testing
> -------
> 
> bin/test.sh
> 
> 
> Thanks,
> 
> Ernie Allen
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message