| There are some real good reasons why one would like the tool.
Assume that I am developing a MOM for Common Agent. I need to do the
entity model, the containment tree, the generate the MSL and stuff that into
MOMGEM to produce the code for the "back end" of the MOM. This all works just
fine for DNA CMIP, but now I want to run my MOM on a system that does not
support CMIP, but only SNMP. Furthermore, I may also want to use a framework
other than MCC (e.g. Sun, HP, or even MSU!). The only use MIB, not MSL
definitions.
In this case, I can either generate the MIB by hand (painfully), or run it
into some tool that does most of the work. In order to make this a reality,
there needs to be some work done on support for datatype mapping and the
MSL needs to be extended to deal with the possible OID structure for various
attributes (or sets of attribute). The current SNMP_OID field in MSL is not
really an acceptable solution.
In reality, the MSL to MIB translation is interesting, but GDMO is also an
issue. Maybe we need "YADL" , yet another descriptor language, which can
encompass the requirements of GDMO, MSL (MCC or NCL flavours), and concise
MIB. Then all we need is back end/front end modules for your favourite
definition language. Of course, fundamental differences in each of the
languages may make "YADL" something that is almost impossible to do. I am
not a "language" or "data representation" person, and would not like to even
guess on the complexity of the problem.
|
| >Furthermore, I may also want to use a framework
>other than MCC (e.g. Sun, HP, or even MSU!). The only use MIB, not MSL
>definitions.
This is our primary reason. We would like to demonstrate our prototype using
both MCC and MSU and wanted to try and sace ourselves some (re)coding effort
as far as the definitions go.
|