[Idr] Should BGP communities be restrictive (Issue #38)?

[Mahesh Jethanandani <[e-mail]>]

The following issue (#38) is documented here <https://github.com/mjethanandani/ietf-bgp-yang/issues/38> for the BGP YANG module.

It reads:

The current community type is defined in a restrictive manner. An example of what that restriction looks like can be seen here <https://github.com/mjethanandani/ietf-bgp-yang/blob/master/src/yang/ietf-bgp-types.yang#L296>. At least one of the observations is that, experience has shown that restricting ranges of communities may have negative impact on things such as policy, which may need to refer to out-of-acceptable-range values. The suggestion is to not restrict these beyond limiting the range to uint32, or what in YANG terms is 0..max.

The question therefore for the WG is, should the YANG model define the community-type in a restrictive manner that will then allow for checking of invalid values, or should it loosen restrictions on the types to allow for an unrestricted range of values to be accepted, knowing that it will fail somewhere deep in the bowels of the system.

An alternative is to define two different types for *-community-type. One that is restrictive and the other that is not. The latter will be used in policy definition, while the former will be used for everything else.

If no strong opinion is heard, the model will maintain the current restrictions on the ranges it has.

Cheers.

Mahesh Jethanandani
[e-mail]