[Ns-bugs] [Bug 822] NetDevice::Mtu attribute not settable by default

[code@nsnam.ece.gatech.edu]

http://www.nsnam.org/bugzilla/show_bug.cgi?id=822





--- Comment #24 from Craig Dowell <craigdo at ee.washington.edu>  2010-04-08 15:23:40 EDT ---
> If we agree that Csma is abstracting many such underlying Ethernet
> technologies and don't want to restrict frame size, then we may not
> have a consistency issue here anymore.

Someone certainly has a consistency issue, it's just not us.

To paraphrase Mathieu, we have models with lots of knobs on them.  In some
(many) cases, it's up to the end user designing the simulation to decide
whether or not a given combination of knobs is meaningful / valid in his or her
scenario.    This may or may not be news to our end users, but it has always
been the case.

I am perfectly fine with this approach, favoring flexibility, but we just need
to document very prominently that the knob settings need to be validated by the
end user, since we aren't going to attempt it -- it will be very easy to
configure a piece of the system that doesn't conform to an underlying spec.  If
a user does something outrageous and impossible, ns-3 may abort at run-time,
but we may go with the user's flow and try to do what he or she says even
though it may mean something unusual is happening.

I support merging the above patch if we include words in appropriate places
about how to deal with inter-related Attributes (i.e., no inter-Attribute
interaction in a setter, and consistency checking only at simulator run time --
when the Attributes are used).  We also need to add words about how you can
easily turn Attribute knobs to get non-standard behavior and it is up to the
user to validate the results (even though that may be quite obvious to some of
us).

I will add an example of knob-turning (the LLC/SNAP switch and super-jumbo
packets in csma seem like good examples) and words of warning in the tutorial.

-- 
Configure bugmail: http://www.nsnam.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.