[db-wg] Proposed changes for object ordering

[Shane Kerr <]

Dear Colleagues,

As mentioned in the posting with the subject "Proposed changes for abuse", 
here is the proposal to change object ordering. As always, please have a 
look and discuss it here.

--
Shane Kerr
Software Manager
RIPE NCC
This is a proposal to display output differently, making recursive
lookup results clearer. 

Currently, the software has a grouping algorithm that sorts objects by
type. For example, if a query finds two INETNUM objects, such as: 

   inetnum:       10.0.0.0 - 10.0.0.127
   .
   .
   admin-c:       PERSON1-RIPE
   tech-c:        PERSON1-RIPE
   .
   .

   inetnum:       10.0.0.128 - 10.0.0.255
   .
   .
   admin-c:       PERSON1-RIPE
   tech-c:        PERSON2-RIPE
   .
   .

For a query, such as: 

$ whois 10.0.0.0 - 10.0.0.0.255 

Here, the related PERSON/ROLE/ORGANISATION objects will also be
returned, after being grouped into types, ending duplication. 

Therefore, the actual result will contain: 

   inetnum:       10.0.0.0 - 10.0.0.127

   inetnum:       10.0.0.128 - 10.0.0.255

   person:        PERSON1-RIPE

   person:        PERSON2-RIPE

This kind of output may be misleading. The user needs to find contact
details by looking first at the "admin-c:" and "tech-c:" attributes,
and then look them up in the PERSON/ROLE objects shown at the end of
the list. This becomes time consuming as the number of objects
returned increases. 

We will change the default output format. We will group objects by
relevance. This change will give the following result for the same
query: 

   % Information related to '10.0.0.0 - 10.0.0.127'

   inetnum:       10.0.0.0 - 10.0.0.127

   person:        PERSON1-RIPE

   % Information related to '10.0.0.128 - 10.0.0.255'

   inetnum:       10.0.0.128 - 10.0.0.255

   person:        PERSON1-RIPE

   person:        PERSON2-RIPE

The changes explained in the section 'Hiding Attributes That Contain
E-Mail Addresses' in the abuse proposal will apply to each grouping in
turn. If only PERSON2-RIPE contains an "abuse-mailbox:" attribute,
PERSON1-RIPE will not be filtered when it first occurs in the output,
but it will be filtered upon its second occurrence. 

The 'traditional' output format of the whois server will still be
possible, by using the '-G' option. 

Note: 

Another proposal could be to present the reordered output if the
'-G' option is passed, otherwise to use the current ordering. This
would leave the average user with the confusing output, and give the
more advanced users the intuitive output.

Making it an option would avoid potential problems with tools
breaking, but some of the abuse changes will break tools so this may
be a good time to change the default.