Hi Tim On 17/12/2015 12:01, Tim Bruijnzeels wrote:
Hi all,
To clarify, we do not create duplicates or overrides.
On 16 Dec 2015, at 16:54, denis <ripedenis@yahoo.co.uk> wrote:
Hi Tim
On 16/12/2015 15:58, Tim Bruijnzeels wrote:
Hi,
To expedite the creation of abuse contacts we've just deployed an enhancement to the RIPE Database web interface.
Whenever you create a new organisation object or you edit an existing one that does not have an abuse contact set, we will display a warning and offer a simple abuse-c creation workflow.
An "abuse-c:" attribute is only required in an ORGANISATION object if it is referenced by a resource object. So the wording in this revised web interface may be a bit confusing to users. It would be better if you do a check on the specified ORGANISATION object and only display this warning if this object should have an "abuse-c:" reference.
From what I understood, warning is not an error and doesn't prevent the creation of the ORGANISATION object.
If you submit the object in the web interface with an empty value for the abuse-c, it will give you the following error: "Please provide an Abuse-c or remove the attribute if you would like to do it later"
So you can explicitly remove the attribute if you want, but we try to guide people into adding the abuse-c because it's a good idea to have it in general (even if it's not strictly required),
Sorry but this is just wrong. The whole design of "abuse-c:" attribute was to use hierarchical inheritance in resource objects. You only put it where it is needed and it is inherited by any more specific objects. It should NOT be duplicated where it is not required. That leads to redundant information being stored in the database. This copy can be forgotten about when the authoritative version referenced from a less specific object is changed and leave invalid data in the database which will override the valid data for some resources.
We are talking here about adding "abuse-c:" to organisation objects, not resource objects.
All inet(6)num and aut-num objects allocated or assigned through the RIPE NCC have an "org:" reference. The referenced organisation must have an "abuse-c:" attribute by policy 2011-06.
If an LIR creates more specific inet(6)num objects, the default behaviour is that the "org:" object are blank, and is the abuse-c 'inherited' from the covering object's "org:". But they can optionally refer to another organisation object. In that case it is also appropriate to have an explicit "abuse-c" on that organisation. In rare cases it may be that two different organisations should use the same abuse contact - this can easily be achieved by referring to the same role object from both organisation objects' "abuse-c:".
In short the duplication that you refer to is not introduced.
You said it yourself below you cannot know why a user is creating an ORGANISATION object. It is quite possible an LIR sub allocates part of an allocation to another company but intends to handle all abuse complaints themselves for their allocation. If you now create an ORGANISATION object for the company holding the sub allocation you tell them they should add an "abuse-c:". Now you are assuming they know they should use the same ROLE object of the LIR and should not change that reference. It is quite likely they will follow your instructions and let you create a new ROLE object for them with their email address. It is far better to encourage users to understand what they are doing and why. If your instructions suggested adding an "abuse-c:" IF required, or even told them to check if they need one, then someone needs to find out if they require it. But to tell them it is missing means they will just follow instructions and add one.
and we often find
that organisation objects are created for sponsored PI or ASN resources without this, and this adds unnecessary overhead to request handling because then we will need to ask for this then.
If you know this is why you are creating this ORGANISATION object then it is reasonable to add an "abuse-c:" at this point. But the wording on your new web form should be clear and suggest adding an "abuse-c:" only IF it is needed.
We are not creating the object. The user is.
That is why your instructions should be clear.
Moreover, during object creation procedure it is unknown what for the ORGANISATION object is created.
You should not be creating objects in the database if you don't know what their immediate use is for. If you do not currently handle any abuse complaints then you should not include an "abuse-c:" attribute in a newly created ORGANISATION object. This is confusing and redundant information and may override the valid contact data.
Again, we are not creating objects. The user is. We cannot know the users full intent at the time they create an organisation object. The duplication you refer to does not exist. And having a dedicated contact method for abuse related matters is best practice. The trash icon and error message in case the value is left blank is indication enough that this is strictly speaking optional.
The "You" did not mean 'you' :) I meant whoever is creating the object. They should know why they are creating it. But since the RIPE NCC doesn't know at this point why they are creating an ORGANISATION object you should not be 'encouraging' them to add an "abuse-c:" at this point. If they do add one of their own and later this is referenced by a sub allocation then you may have overridden the LIRs abuse contact details. My point is simple. People should know what they are doing. An "abuse-c:" attribute should be added where and when needed. Telling people to add it regardless, with the possibility of the wrong data being used if the blindly follow your instructions, is not useful. cheers denis
Cheers Tim
cheers denis
Indeed, we don't know this at creation time.
Regards,
Tim
All the best, Piotr
-- gucio -> Piotr Strzyżewski E-mail: Piotr.Strzyzewski@polsl.pl