Shepherd writeup

This document is proposed for publication on the Independent Stream as an Informational RFC.

The document describes a common command set to provide application programmers uniform access to the Looking Glass services.

The Introduction makes clear that "This work is not the output of the IETF and is presented in the hope that Looking Glass implementers will offer a common programmable interface."

The ISE was first approached about this document in Jun 2016. Since then it has taken a lot of effort to attract expert reviews, but two were finally obtained from Geof Huston (rather brief) and John Heasley (included below). This barely passes my criteria for publication, but the document had already been committed for publication before I was appointed as ISE.

I performed my own review of -08 and Markus took two revisions to address my comments.

=== John Heasley review

1) There are a few idiomatic language points that I might change.  For
example, "The client MUST NOT be able to execute random commands on the
targeting router."; targeting -> targeted or target.  Clearly these
only affect the fluidity of reading.

2) The document mentions caching in terms of intermediaries, such as
proxies, caching responses.  It is also common for a LG to cache some
responses of commands that are time consuming or burdening for devices to
run.  Clearly the LG could convey that a response is from the cache in the
performed attribute of a successful response, but perhaps the
document should address this possibility.

3) I believe it would be useful to propose a format for json encoding of
the data portion of a successful response.  Raw format is only useful for
presentation to humans, but a standard or even suggested format would be
useful for data collection.  I fully understand that this is a rat bag,
but not every field need be defined.  The draft author has an example
implementation; perhaps he has done some of this there already and that
might be sufficient.  One could also take this probably verbatim from
yang models - or just suggest that.