Re: MODE_SET_*

Dave Shield wrote:

>> Why don't we have one and
>>only MODE_SET state (like in GET or GETNEXT)?
> 
> Because changing settings is a lot more complex than
> simply reporting them.  Particularly ensuring that
> either everything gets changed, or nothing does.
> (The "as if simultaneously" behaviour).

Now I understand it. I missed the part that "a SET request can set
several variables, not only one". Then it is some kind of attempt to
reach a pseudo-atomic operation (which never gets really true since
COMMIT may unexpectedly fail).

>>Really I only need to implement a simple SET: my program should check
>>whether received OID is the right one, then store the value and exit.
> 
> But what if the agent is handling some other, more complex assignment
> as part of the same request.  Just because your particular SET handling
> is (apparently) simple, doesn't mean that the whole agent can be so
> forgiving.

I was only thinking about setting a single variable.

>>I've thought of implementing all of this in one state (for instance,
>>MODE_SET_RESERVE1), but I don't know if that's not good...
> 
> 
> Typically, you should check the requested value is OK in the RESERVE1
> pass.  If you are *certain* that the assignment can't possibly fail,
> then it's safe to do this in the COMMIT pass.  Otherwise, you should
> do it in the ACTION pass, and be ready to reverse it in the UNDO pass.

Well, actually I'm using RESERVE1 and ACTION states. But since my write
operation is not "comittable"/"undoable" perhaps I'll change ACTION by
COMMIT (it really doesn't affect my program's behaviour).

Thanks for your response. It has been very useful :)

Regards,
-Román


-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
Net-snmp-users mailing list
[email protected]
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users