[db-wg] NWI-4 - role of status: field in multivalued status context - reprise

Hi guys

Just out of interest why do people think having a dual attribute
primary key pair is better than allowing multiple "status:" attributes
in very specific cases?

Multiple status attributes is probably far simpler to implement and
maybe easier to parse. Changing the pkey affects the way millions of
objects are handled. Adding a second status affects (tens of)
thousands of objects. In both cases multiple objects may be returned
in a query for address space which all existing user scripts will have
to accomodate. Changing the pkey will require more business rules to
(dis)allow certain status combinations. Unless it is OK to have an
ALLOCATED PA, :LIR-PARTITIONED PA, SUB-ALLOCATED PA, ASSIGNED PA all
with the same prefix. Changing the pkey also means changes to the
query code to allow a query for a specific object using the
prefix/status pair or a more general query on just the prefix to
return possibly multiple objects.

cheers
denis
co-chair DB-WG

On Tue, 8 Jun 2021 at 16:22, Cynthia Revström via db-wg <db-wg at ripe.net> wrote:
>
> Hi,
>
> I would just like to say that I would also like this to be implemented if the DB team determines it to be feasible. :)
>
> -Cynthia
>
> On Tue, Jun 8, 2021 at 1:31 PM Edward Shryane via db-wg <db-wg at ripe.net> wrote:
>>
>> Hi Carsten,
>>
>> The DB team will create a proof of concept to see if this is feasible, and I will report our findings to you and the DB-WG.
>>
>> Regards
>> Ed Shryane
>> RIPE NCC
>>
>>
>> > On 8 Jun 2021, at 12:03, Carsten Schiefner via db-wg <db-wg at ripe.net> wrote:
>> >
>> > Dear all,
>> >
>> > maybe someone from the NCC would feel inclined to comment on the
>> > feasibility of the below?
>> >
>> > Thanks & best,
>> >
>> >       -C.
>> >
>> > On 21.05.2021 12:55, Piotr Strzyzewski via db-wg wrote:
>> >> On Fri, May 21, 2021 at 12:45:35PM +0200, Carsten Schiefner via db-wg wrote:
>> >>
>> >> Hi Carsten,
>> >>
>> >>> after a quick chat with Dennis on this, he encouraged me to toss this
>> >>> into the seemingly stalled, but not yet dead debate:
>> >>>
>> >>> Right now, only the "inet[6]num:" attribute is the primary key to, of or
>> >>> for inet[6]num objects.
>> >>>
>> >>> I wonder if it would be possible to make the tuple
>> >>> ("inet[6]num:","status:") the primary key instead: that should solve the
>> >>> challenge to have an assignment that shall have the size of an allocation.
>> >>
>> >> This is very good idea.
>> >>
>> >>> In case this has been exhaustedly discussed here already, please excuse
>> >>> my ignorance - I then obviously have failed to spot the respective
>> >>> contribution in the archives.
>> >>
>> >> I do not recall such discussion.
>> >>
>> >> Best,
>> >> Piotr
>> >
>>
>>