Re: ConstrainableLookupLocator identity

2018-10-18 Thread Peter Firmstone

Hmm,

I should have said constraints shouldn't be part of LookupLocator 
identity below, my bad.   Perhaps if ConstrainableLookupLocator used 
decoration, rather than inheritance, the design intent would have been 
clearer.  I don't like the fact that a ConstrainableLookupLocator can be 
considered equal to a LookupLocator, but it breaks the following tests.  
So I'm going to lodge an issue and back it out.   I think the design 
intent is like decorating a LookupLocator with constraints, it's was 
never intended to change it's identity.


org/apache/river/test/impl/mahalo/AdminIFTest.td
org/apache/river/test/impl/mahalo/AdminIFShutdownTest.td

N.B. This only affects people using secure services.

Regards,

Peter.

On 18/10/2018 6:40 AM, Bryan Thompson wrote:

I've never used that aspect.  Nothing to offer.
B
On Wed, Oct 17, 2018 at 3:57 AM Peter Firmstone
  wrote:

LookupLocator's identity contract:

/**
   * Two locators are equal if they have the samehost  and
   *port  fields. The case of thehost  is
ignored.
   * Alternative forms of the same IPv6 addresses for the
host
   * value are treated as being unequal.
   */

At some point in history, here,
http://svn.apache.org/viewvc/river/jtsk/trunk/src/net/jini/discovery/ConstrainableLookupLocator.java?r1=1034266=1034267;

FindBug's identified the ConstrainableLookupLocator didn't override
equals, so I implemented it, however while it seemed to make sense at
the time to include constraints, I'm finding that it's causing problems
for discovery magement and now I'm thinking that the constraints,
probably shouldn't be part of constraints.

What are your thoughts?

Regards,

Peter.




Re: ConstrainableLookupLocator identity

2018-10-17 Thread Bryan Thompson
I've never used that aspect.  Nothing to offer.
B
On Wed, Oct 17, 2018 at 3:57 AM Peter Firmstone
 wrote:
>
> LookupLocator's identity contract:
>
> /**
>   * Two locators are equal if they have the same host and
>   * port fields. The case of the host is
> ignored.
>   * Alternative forms of the same IPv6 addresses for the
> host
>   * value are treated as being unequal.
>   */
>
> At some point in history, here,
> http://svn.apache.org/viewvc/river/jtsk/trunk/src/net/jini/discovery/ConstrainableLookupLocator.java?r1=1034266=1034267;
>
> FindBug's identified the ConstrainableLookupLocator didn't override
> equals, so I implemented it, however while it seemed to make sense at
> the time to include constraints, I'm finding that it's causing problems
> for discovery magement and now I'm thinking that the constraints,
> probably shouldn't be part of constraints.
>
> What are your thoughts?
>
> Regards,
>
> Peter.