Re: [netmod] 6991bis: address-with-prefix-length

[Christian Hopps <[e-mail]>]

> On Apr 2, 2019, at 22:16, Alex Campbell <[e-mail]> wrote:
> 
> I don't think this is joining an address with a prefix - this is joining an address with a prefix *length*.
> If it was two separate values, it would still be constraining the address to be covered by the prefix, since there would be no way to define a prefix that the address didn't cover.

It is joining the two things as far as what data it *represents* (i.e., how it can be used). You are correct if we are just talking about the physical bits being stored, but semantics count. Otherwise I think you are agreeing with me that by representing the 2 values this way it necessarily constrains their values the way we want it to.

Thanks,
Chris.

> 
> Also a distinction from ipv4-prefix is that an ipv4-prefix is completely meaningless if either part is removed, whereas an ip-address-and-prefix-length still specifies an IP address if you remove the prefix length.
> 
> Alex
> 
> ________________________________________
> From: netmod <[e-mail]> on behalf of Christian Hopps <[e-mail]>
> Sent: Wednesday, 3 April 2019 7:52 a.m.
> To: Martin Bjorklund
> Cc: [e-mail]
> Subject: Re: [netmod] 6991bis: address-with-prefix-length
> 
>> On Apr 2, 2019, at 2:27 PM, Martin Bjorklund <[e-mail]> wrote:
>> 
>> "Rob Wilton (rwilton)" <[e-mail]> wrote:
>>> Hi Acee,
>>> 
>>> Having re-read the thread, I think that "ip-address-prefix" is going
>>> to be confusing, since I had incorrectly assumed that the type being
>>> defined was an IP prefix, but as you have pointed out there is already
>>> a type for that.
>>> 
>>> I think that we need to choose this name very carefully or otherwise I
>>> suspect that folks will accidentally use the wrong type.
>>> 
>>> So having the type as "ip-address-and-prefix-length" or
>>> "ip-addr-and-prefix-len" now seems like a clearer choice to me.
>> 
>> The combined type really specifies (i) an ip address and (ii) an ip
>> prefix.  The prefix happens to be specified with a length indicator.
>> So I think the name "ip-address-and-prefix" is the correct one.
>> 
>>> I
>>> think that I also now agree with Martin that this is really merging
>>> two values into one leaf.
>> 
>> And for the record (again, perhaps), I think this is a bad idea in
>> general, and I am not sure an exception is needed in this case.
> 
> (Again :) this is not just joining two values, it is also constraining the address value to be a value covered by the prefix. The common use case is for interface state where the interface has an address which should be within the prefix assigned to the network the interface attaches to.
> 
> This isn't just about saving space.
> 
> I also agree that "-and-" is a really good idea, saving a few characters in an identifier when doing so has shown to cause the identifier to be misunderstood is not an optimization IMO.
> 
> Thanks,
> Chris.
> 
>> 
>> 
>> /martin
>> 
>> 
>>> Where is this type going to be used?  Is it only used for configuring
>>> host address/prefix?  Or are there other uses cases as well?
>>> 
>>> Thanks,
>>> Rob
>>> 
>>> 
>>>> -----Original Message-----
>>>> From: Acee Lindem (acee) <[e-mail]>
>>>> Sent: 02 April 2019 16:52
>>>> To: Rob Wilton (rwilton) <[e-mail]>; Martin Bjorklund
>>>> <mbj@tail-
>>>> f.com>; [e-mail]
>>>> Cc: [e-mail]
>>>> Subject: Re: [netmod] 6991bis: address-with-prefix-length
>>>> 
>>>> Hi Rob,
>>>> 
>>>> On 4/2/19, 11:37 AM, "netmod on behalf of Rob Wilton (rwilton)"
>>>> <netmod-
>>>> [e-mail] on behalf of [e-mail]> wrote:
>>>> 
>>>> 
>>>> 
>>>>> -----Original Message-----
>>>>> From: netmod <[e-mail]> On Behalf Of Martin
>>>> Bjorklund
>>>>> Sent: 02 April 2019 13:47
>>>>> To: [e-mail]
>>>>> Cc: [e-mail]
>>>>> Subject: Re: [netmod] 6991bis: address-with-prefix-length
>>>>> 
>>>>> Juergen Schoenwaelder <[e-mail]> wrote:
>>>>>> If you go back ~20 messages, my proposal was ip-address-prefix,
>>>>>> ipv4-address-prefix, and ipv6-address-prefix.
>>>>> 
>>>>> Do we agree that this type really specifies two values in one?  If so
>>>>> I think
>>>> the
>>>>> "and" is useful.
>>>> 
>>>>   Isn't an "IP prefix" made up of an "IP address" and a "prefix length"?
>>>> 
>>>> This was my confusion as well since the ipv4-prefix and ipv6-prefix
>>>> types
>>>> (ietf-inet-types) have been used when they probably shouldn't have
>>>> been.
>>>> Note that they both have the constraint of not allowing the host bits
>>>> to be set
>>>> should they should be used in situations like interface address
>>>> assignment.
>>>> 
>>>> Excerpted from RFC6991 ipv4-type definition (note the last sentence):
>>>>    description
>>>>       "The ipv4-prefix type represents an IPv4 address prefix.
>>>>        The prefix length is given by the number following the
>>>>        slash character and must be less than or equal to 32.
>>>> 
>>>>        A prefix length value of n corresponds to an IP address
>>>>        mask that has n contiguous 1-bits from the most
>>>>        significant bit (MSB) and all other bits set to 0.
>>>> 
>>>>        The canonical format of an IPv4 prefix has all bits of
>>>>        the IPv4 address set to zero that are not part of the
>>>>        IPv4 prefix.";
>>>> 
>>>>   So, I think that the names above are probably right, or otherwise if
>>>>   you
>>>> want the "and" then perhaps it should be
>>>> "ip-address-and-prefix-length" -
>>>> which seems clunky?
>>>> 
>>>> I think the original suggestion of ipxx-address-prefix would be ok.
>>>> 
>>>> Thanks,
>>>> Acee
>>>> 
>>>>   Thanks,
>>>>   Rob
>>>> 
>>>> 
>>>>> 
>>>>> Also note that the current text in RFC 6991 says:
>>>>> 
>>>>>    The ipv4-prefix type represents an IPv4 address prefix.
>>>>> 
>>>>> so having a type ipv4-address-prefix for something that is not (only)
>>>>> an
>>>>> "ipv4 address prefix" is imo confusing.
>>>>> 
>>>>> 
>>>>> /martin
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>>> 
>>>>>> /js
>>>>>> 
>>>>>> On Tue, Apr 02, 2019 at 11:13:09AM +0000, tom petch wrote:
>>>>>>> ----- Original Message -----
>>>>>>> From: "Jeff Tantsura" <[e-mail]>
>>>>>>> To: <[e-mail]>; "Kristian Larsson" <[e-mail]>
>>>>>>> Sent: Monday, April 01, 2019 11:09 PM
>>>>>>> 
>>>>>>> What Kristian has proposed makes sense, in favor.
>>>>>>> 
>>>>>>> <tp>
>>>>>>> 
>>>>>>> Yes, I support this idea and we should be able to come up with a
>>>>>>> more user-friendly name;  address-prefix or address-length ?
>>>>>>> 
>>>>>>> Tom Petch
>>>>>>> 
>>>>>>> p.s.
>>>>>>> 
>>>>>>>  identifier          = (ALPHA / "_")
>>>>>>>                        *(ALPHA / DIGIT / "_" / "-" / ".")
>>>>>>> 
>>>>>>> Cheers,
>>>>>>> Jeff
>>>>>>> On Apr 1, 2019, 1:09 PM -0700, Kristian Larsson
>>>>>>> <[e-mail]>, wrote:
>>>>>>>> Hello Mahesh,
>>>>>>>> 
>>>>>>>> On 2019-04-01 21:40, Mahesh Jethanandani wrote:
>>>>>>>>> 
>>>>>>>>>> On Apr 1, 2019, at 10:29 AM, Martin Bjorklund <mbj@tail-
>>>> f.com>
>>>>>>> wrote:
>>>>>>>>>> 
>>>>>>>>>> I know that this type is convenient, esp. if you use it for
>>>>>>>>>> manual input, but I wonder if it really is good practice to
>>>>>>>>>> squeeze two values into one.
>>>>>>>>> 
>>>>>>>>> Agree. The combination makes sense for CLI, but for modeling
>>>>>>>>> the
>>>>>>> address and prefix should be separate.
>>>>>>>> 
>>>>>>>> Okay, then why do we have an ip-prefix data type at all? With the
>>>>>>>> same line of argument you apply, it should be split up.
>>>>>>>> 
>>>>>>>> So you're the third person bringing up CLI. I don't get this at
>>>>>>>> all. I don't see how CLI are different from everything else. This
>>>>>>>> is about
>>>>>>> data
>>>>>>>> modeling and data modeling is about expressing the world in a
>>>>>>>> data
>>>>>>>> modeling language. It's like painting a picture but instead of a
>>>>>>>> brush you have a schema language like YANG. What do you see?
>>>>>>>> Express it. It doesn't matter if the purpose is a CLI, a web page
>>>>>>>> or just exposing it via NETCONF for another system to consume.
>>>>>>>> 
>>>>>>>> I think address-and-prefix-length is natural. JUNOS uses this
>>>>>>>> format.
>>>>>>> XR
>>>>>>>> uses this format (for IPv6 at least). Nokia SROS uses this
>>>>>>>> format.
>>>>>>>> 
>>>>>>>> We have written a bunch of models where the lack of this IMHO
>>>>>>>> makes
>>>>>>> them
>>>>>>>> less elegant. I'd like for there to be an IETF standard data type
>>>>>>>> to make those models more elegant.
>>>>>>>> 
>>>>>>>> Kind regards,
>>>>>>>> Kristian.
>>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> netmod mailing list
>>>>>>>> [e-mail]
>>>>>>>> https://www.ietf.org/mailman/listinfo/netmod
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>> --------------------------------------------------------------------
>>>>>>> ----
>>>>>>> --------
>>>>>>> 
>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> netmod mailing list
>>>>>>>> [e-mail]
>>>>>>>> https://www.ietf.org/mailman/listinfo/netmod
>>>>>>>> 
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> netmod mailing list
>>>>>>> [e-mail]
>>>>>>> https://www.ietf.org/mailman/listinfo/netmod
>>>>>> 
>>>>>> --
>>>>>> Juergen Schoenwaelder           Jacobs University Bremen gGmbH
>>>>>> Phone: [phone]         Campus Ring 1 | 28759 Bremen |
>>>> Germany
>>>>>> Fax:   [phone]         <https://www.jacobs-university.de/>
>>>>>> 
>>>>>> _______________________________________________
>>>>>> netmod mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/netmod
>>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> netmod mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/netmod
>>>> 
>>>>   _______________________________________________
>>>>   netmod mailing list
>>>>   [e-mail]
>>>>   https://www.ietf.org/mailman/listinfo/netmod
>>>> 
>>>> 
>>> 
>> _______________________________________________
>> netmod mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/netmod
>