Hi Ren, After a scan of the mailing list archives for the srvloc project on sf.net, I found the following message submitted by Matt Peterson:
http://sourceforge.net/mailarchive/message.php?msg_id=3209418 This message explains the rationale behind disabling incremental service registration and deregistration. I agree with Matt's assessment and feel that we should keep the code as is - incremental service registration is not supported in OpenSLP because using it overtaxes the SLP protocol. If you need incremental registration and deregistration, perhaps you should consider using LDAP instead of SLP. Any comments are appreciated (from anyone). John > -----Original Message----- > From: John Calcote [mailto:john.calc...@gmail.com] > Sent: Tuesday, January 08, 2013 11:09 AM > To: 'Wang, Ren' > Cc: OpenSLP Devel Mailing List (openslp-devel@lists.sourceforge.net) > Subject: RE: SLPReg fresh=false > > (Adding devel list back in so others can chime in if they have input) > > Hi Ren, > > Ok - I was correct in my understanding then - I thought I understood that you > wanted incremental registrations. My original reply to you was that the > entire concept of incremental registrations appears to be deprecated in > slpv2bis, which is the standard that OpenSLP is trying to follow. > > Incremental registrations is controlled by the FRESH flag in SLP message > headers, and the FRESH flag is required to be set to 1 by slpv2bis. What I > *don't* know is why. I don't see any explanation anywhere of why this flag > was deprecated and required to be set to 1 in message headers. I presume > that Matt Peterson disabled the use of the boolean fresh field in the SLPReg > api in order to support the deprecation of incremental registrations. > > In this document: http://srvloc.sourceforge.net/compatibility.html the fresh > flag is listed under the SLPv2 column as: > > "When this flag is present in a SrvReg, this registration overwrites any > existing registration with the same URL. When this flag is absent, a SrvReg will > incrementally add to an existing registration." > > And under the slpv2bis column as: > > "As RFC 2608, except that the Fresh Flag MUST be set on registrations. If not, > return a FRESH_MUST_BE_SET error?" (The error code to be returned was > properly defined after this document was created.) > > In other words, since the current implementation of OpenSLP tries to > support SLPv2bis as closely as possible, we've disabled incremental > registration by ignoring the Boolean fresh argument passed to SLPReg and > hard-coding the FRESH flag in the SrvReg message header to 1. Note that this > flag is not a tri-state - the field is always present, and must be either 1 or 0. At > certain places in the documents referenced on this thread, it appears that > the flag may be present or not, and if present it must be 1 and may not be > zero. The flags word is always present, and the FRESH flag is hard-coded to a > particular position in this word, so it must be present, and must be set to 1. > Since setting this flag to 1 means the registration is fresh, the registration will > overwrite any existing registration. > > Once again, I don't know why this was done - no documents I've been able to > find on the topic seem to indicate the rationale or discussion of the issue that > caused the change. If anyone on the list knows, please chime in. > > Please understand Ren, that I'm not against incremental refresh - if I > understood the rationale begin removing it, I would be able to make an > intelligent decision about whether to follow the standard in this area. Since I > don't know why it was deprecated, I have only the wording of the standard > to go by. If you can find any documentation on the net as to why it was > removed in the first place, I'd appreciate your insight. > > John > > > -----Original Message----- > > From: Wang, Ren [mailto:ren.w...@nuance.com] > > Sent: Tuesday, January 08, 2013 9:21 AM > > To: John Calcote > > Subject: RE: SLPReg fresh=false > > > > Hi John, > > > > What we are looking for is to support incremental service registrations. > > > > For example, if there is a service registered with attribute (user_id= > > Ren), and later a new user added to the service, so the increment > > registration will call SLPReg with attr (user_id=John) and fresh flag > > set to false to indicate it is an incremental registration. In the > > registry, the service should have attribute (user_id=Ren, John). > > > > > > > > > > > > -----Original Message----- > > From: John Calcote [mailto:john.calc...@gmail.com] > > Sent: Tuesday, January 08, 2013 11:13 AM > > To: Wang, Ren > > Subject: RE: SLPReg fresh=false > > > > I'm sorry Ren, I still don't understand what you're after. Please > > forgive my incomprehension - if you could explain exactly what you > > want to use the fresh flag for and why, then perhaps I'd understand > > what you're asking. I was simply explaining why it's currently implemented > (or not) the way it is. > > > > John > > > > > -----Original Message----- > > > From: Wang, Ren [mailto:ren.w...@nuance.com] > > > Sent: Tuesday, January 08, 2013 8:07 AM > > > To: john.calc...@gmail.com > > > Subject: RE: SLPReg fresh=false > > > > > > Hi John, > > > > > > Thank you again for your response and the URLs. > > > > > > Maybe my question was not clear to you, but I was trying to ask if > > > OpenSLP will support fresh=false instead of not set the fresh flag. > > > On you second > > URL, > > > page 6, it says "FRESH" MUST be set to 1 on every SrvReg. > > > Otherwise, > > MUST > > > be 0." > > > > > > Since current OpenSLP implementation does not support 0 for SrvReg. > > > Based on the OpenSLP.org, "Currently, OpenSLP does not support > > > incremental registrations. If fresh is SLP_FALSE, SLPReg() will > > > return SLP_NOT_IMPLEMENTED." > > > > > > This is why I want to know if you plan to support it. > > > > > > Regards, > > > > > > Ren > > > > > > -----Original Message----- > > > From: John Calcote [mailto:john.calc...@gmail.com] > > > Sent: Monday, January 07, 2013 6:37 PM > > > To: Wang, Ren > > > Cc: openslp-us...@lists.sourceforge.net > > > Subject: RE: SLPReg fresh=false > > > > > > Hi Ren, > > > > > > The FRESH flag was deprecated after RFC 2608 was published. > > > > > > See: > > > > > > http://srvloc.sourceforge.net/new_drafts/draft-guttman-svrloc-as-00. > > > tx > > > t > > > http://srvloc.sourceforge.net/new_drafts/draft-guttman-svrloc-rfc260 > > > 8b > > > is- > > > 01. > > > txt > > > > > > In the first document it states on page 3 that an error > > > (INVALID_UPDATE) > > is > > > returned by the SA/DA for registrations that don't set the FRESH > > > flag in > > post > > > slpv2 implementations (slpv2bis - the second document - pp 6, 7, 21). > > > The slpv2bis document isn't clear as to why the FRESH flag must be > > > set > > > - > > just > > > states that it must be set. I presume it's a security issue of some kind. > > > > > > John > > > > > > > -----Original Message----- > > > > From: Wang, Ren [mailto:ren.w...@nuance.com] > > > > Sent: Monday, January 07, 2013 9:08 AM > > > > To: John Calcote > > > > Cc: openslp-us...@lists.sourceforge.net > > > > Subject: SLPReg fresh=false > > > > > > > > Hi John, > > > > > > > > Is there a plan to support fresh=false for SLPReg API? > > > > > > > > Since it is a required feature for our project, we may need to > > > > provide the change to the OpenSLP if there is no short term plan > > > > to > > support it. > > > > > > > > Regards, > > > > > > > > Ren > > ------------------------------------------------------------------------------ Master SQL Server Development, Administration, T-SQL, SSAS, SSIS, SSRS and more. Get SQL Server skills now (including 2012) with LearnDevNow - 200+ hours of step-by-step video tutorials by Microsoft MVPs and experts. SALE $99.99 this month only - learn more at: http://p.sf.net/sfu/learnmore_122512 _______________________________________________ Openslp-devel mailing list Openslp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openslp-devel
