Most of NML is solid, and has been discussed and tested long enough that I see 
no problems in pushing it as a standard.

However, there are still some features, which I consider useful but still 
experimental in the sense that they have not been thoroughly discussed.

Proposal 1: We identify these features, and mark them as "EXPERIMENTAL", meaning
 that they MUST be accepted as syntactically valid NML, but MAY be ignored and 
are NOT RECOMMENDED for use in production services.

Proposal 2: We describe these features in the base document, rather than in an 
extension document.

Proposal 3: It concerns the following features:
  - artf6512: isParallelCompound
  - artf6579: LinkGroups allowed in <Link> isSerialCompoundLink <Ordered List of
 (Link|LinkGroup)s>
  - artf6507: hasChannel
  - artf6569 (bullet 7): PortGroup --hasService--> AdaptationService
  - artf6571: source/destination labels
  - artf6580: Allow Node|Topology --(isSource|isSink)--> Link

Most of NML is solid, and has been discussed and tested long enough that I see 
no problems in pushing it as a standard.

However, there are still some features, which I consider useful but still 
experimental in the sense that they have not been thoroughly discussed.

Proposal 1: We identify these features, and mark them as "EXPERIMENTAL", meaning
 that they MUST be accepted as syntactically valid NML, but MAY be ignored and 
are NOT RECOMMENDED for use in production services.

Proposal 2: We describe these features in an extension document, rather than in 
the base document.

Proposal 3: It concerns the following features:
  - artf6512: isParallelCompound
  - artf6579: LinkGroups allowed in <Link> isSerialCompoundLink <Ordered List of
 (Link|LinkGroup)s>
  - artf6507: hasChannel
  - artf6569 (bullet 7): PortGroup --hasService--> AdaptationService
  - artf6571: source/destination labels
  - artf6580: Allow Node|Topology --(isSource|isSink)--> Link