Re: event subscribe operation SET-PROPERTY ?
Jamal Hadi Salim <[email protected]>
| Newsgroups | gmane.ietf.forces |
|---|---|
| Organization | ZNYX Networks |
| Message-ID | <1161375152.5201.23.camel@jzny2> |
Weiming, On Tue, 2006-17-10 at 13:59 +0800, Wang,Weiming wrote: > Jamal, > > ----- Original Message ----- > From: "Jamal Hadi Salim" <[email protected]> > > > Weiming, > > > > 1) I would rather we dont go back and rehash the accessibility of > > properties at this time for the sake of time to get the RFC review. > > IMO, we screwed up by not having it in the protocol because it was > > discussed on the concalls and list - so thanks very much to Fenggen for > > catching it. > > To fix this i have opened an issue (#88) in the tracker so that it is > > fixed in the draft. In retrospect thats what we should have done with > > the properties (opened tracker issues) > do you mean to fix it in the protocol draft or in the model draft? I > suppose the protocol draft still leaves open chances to fix some thing > if we think it is necessary. I just think the SET-PROPERTY is not necessary > and we may be able to adopt other choice. But we had many discussions on the concals and Joel did a lot of work putting together the thoughts in writing that were posted on the list. I dont recalll you in particular saying anything - and often times silence is implied consent at the IETF. > > > > 2) On resource usage that you point out below: > > IMO, one could live with threshold, eventHysteresis,eventHysteresis > > being always present because they are as many of those as there are > > events. And i dont see a gazillion events anywhere. The nasty one maybe > > the 8 bits defined for base properties since that is present in every > > element. > It still appears awkward some for some events reflecting status changes only. I agree. It also depends on implementation approach; you could ease the use. However, for the reasons i described above - mainly being time to get the RFC out i think we should let this go. If you feel very strongly about it, then by all means lets re-open the discussion; we still have not implemented this feature, so I have no implementation experience to share. cheers, jamal