> >  > On behalf of the RBridges team I am requesting feedback
 > >  > from the community on our plan to introduce support for
 > >  > simulated links in OpenSolaris. The document included
 > >  > in-line below introduces simlinks and is in the form of 
 > >  > a PSARC fast-track document.
 > > 
 > > I like this proposal; it seems very clean and fits in well with the dladm
 > > architecture.   A few questions:
 > > 
 > >    * I see this case is Committed, so I presume we planning to
 > >      document this construct in dladm(1M)?  Or do we feel that's
 > >      not worthwhile?
 > 
 > I feel it should be left out of dladm man page until we gain more 
 > experience with simlinks. I can create a page on opensolaris that could 
 > list the commands and private properties.

The Interface Taxonomy document has this to say about Committed:

  We publish the specification of these interfaces, typically as manual
  pages or other product documentation.

I guess what you propose above could be "other product documentation", but
it feels a bit loose.

 > >    * Related to the previous question: must the other endpoint be a
 > >      simlink?
 > 
 > Yes it must be simlink. It is possible to remove this restriction 
 > to meet a testing need in future.

Would this also remove the requirement that peers be symmetric?  Or would
e.g., a phys link with a peer of a simlink be able to pull packets both
from the peer simlink and the wire?

 > >    * Just a nit, but when I hear the word "simlink" I instantly think
 > >      of "symlink".  Also, none of our other link classes have "link"
 > >      in the name.
 > 
 > Agreed though the geeky name might work here given the use case.

OK.

 > >    * It seems like some of the stuff shown by show-simlink is only
 > >      due to deficiences with show-link.  In particular, it's hard
 > >      to understand why show-link doesn't include the hardware address
 > >      and it should probably also have the media type.
 > 
 > True but given that we already have show-* commands for other link classes
 > I felt it would be expected there be a show-simlink command.

To be clear, my issue wasn't with show-simlink, just that ultimately the
various show-* commands focused on displaying information that was unique
to their domain rather than being forced into displaying general
information because of deficiencies in other show-* commands (see 6802707
for more on this).  To that end, I'd expect we'd revise this whenever
someone fixes up show-link to be more complete.

-- 
meem
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to