Let's be clear that there are two issues, in actuality unrelated, that got "combined" to try to understand the original guidance:
1) Should link annotations (attributes on the link itself) have any information "derived" from the target resource; for example, should you "title" the link with the title of the resource? Jim, Arthur, and others (including me if I had responded) would say "NO" because whatever that information is when you start, it's likely to change and then the annotation on the link is misleading. 2) Should a client cache a resource? Maybe--it may be up to the client to care about whether it's fresh, or some standard HTTP constructs could be used to tell if it's ok to use the cached copy. The idea of caching came into this thread as a possibly related to the "meaning" behind the link annotation, but any discussion of caching has to be on its own merits, not related to link annotations (especially if the "NO" is heeded for (1)). Now, question--Jim you say that e-tag probably won't work. Why not? Are you thinking about the issue that the OSLC resource may not in actuality be a single file on the server, so how does the server maintain a meaningful e-tag? That was discussed at one point in the AM workgroup I know. Has that been further discussed? It seems like an issue that's both important and hard. Andy Berner Lead Architect, ISV Technical Enablement and Strategy IBM Rational Business Development 972 561-6599 [email protected] Ready for IBM Rational software partner program - http://www.ibm.com/isv/rational/readyfor.html |------------> | From: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |James Conallen/Philadelphia/IBM@IBMUS | >--------------------------------------------------------------------------------------------------------------------------------------------------| |------------> | To: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |Arthur Ryman <[email protected]> | >--------------------------------------------------------------------------------------------------------------------------------------------------| |------------> | Cc: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |[email protected], [email protected], [email protected] | >--------------------------------------------------------------------------------------------------------------------------------------------------| |------------> | Date: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |09/01/2010 08:55 AM | >--------------------------------------------------------------------------------------------------------------------------------------------------| |------------> | Subject: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |Re: [oslc-core] Provided guidance for adding relationship labels | >--------------------------------------------------------------------------------------------------------------------------------------------------| |------------> | Sent by: | |------------> >--------------------------------------------------------------------------------------------------------------------------------------------------| |[email protected] | >--------------------------------------------------------------------------------------------------------------------------------------------------| Good point. I wonder if it is then appropriate to provide some guidance on the use HTTP cache control - i don't think etags will help here. So, if a provider wants to cache some information about the object resource, like the title (using Steve's example), that server should also look for an Cache-Control header and then decide if it can cache the information, and if it can, how long it can cache it. (this information can also be stored in the reified statement). while all this can be done, and presumably to address expected performance issues, i still think the guidance should be to refrain from using link annotations to describe information about the target (object), and instead only use it to describe the link itself. <jim/> jim conallen CAM Lead Architect [email protected] Rational Software, IBM Software Group (Embedded image moved to file: pic18631.gif)Inactive hide details for Arthur Ryman ---08/31/2010 04:19:13 PM---Steve/James, There are standard HTTP practices that describeArthur Ryman ---08/31/2010 04:19:13 PM---Steve/James, There are standard HTTP practices that describe how to safely cache From: Arthur Ryman <[email protected]> To: James Conallen/Philadelphia/IBM@IBMUS Cc: [email protected], [email protected], [email protected], Steve K Speicher/Raleigh/IBM@IBMUS Date: 08/31/2010 04:19 PM Subject: Re: [oslc-core] Provided guidance for adding relationship labels Steve/James, There are standard HTTP practices that describe how to safely cache resources, and well-behaved clients (e.g. Web browsers) should use those. OSLC clients should be well-behaved HTTP clients. This includes the use of ETags, expiry dates, and HEAD requests. Regards, ___________________________________________________________________________ Arthur Ryman, PhD, DE Chief Architect, Project and Portfolio Management IBM Software, Rational Markham, ON, Canada | Office: 905-413-3077, Cell: 416-939-5063 Twitter | Facebook | YouTube From: James Conallen <[email protected]> To: Steve K Speicher <[email protected]> Cc: [email protected], [email protected] Date: 08/31/2010 03:29 PM Subject: Re: [oslc-core] Provided guidance for adding relationship labels Sent by: [email protected] my 2c, I worry about guidance that that suggests that it is ok to essentially cache information about a resource that is being referenced (and managed by) on another server. If this is to be a practice, what are the recommendations for ensuring that this information remains in sync. Looking at the referenced example, what happens if the owner of the resource 123 changes its title to "Enhancement 123: Enable multi-root installs"? Will this have to be manually updated? If not, does the system automatically update properties of links whenever it detects them. While I do recognize this may be a way to save a GET call. I don't think it represents a best practice. <jim/> jim conallen CAM Lead Architect [email protected] Rational Software, IBM Software Group Steve K Speicher---08/31/2010 02:47:42 PM---I wanted to call out some specification updates that was created for handling of relationship label From: Steve K Speicher/Raleigh/IBM@IBMUS To: [email protected] Cc: [email protected] Date: 08/31/2010 02:47 PM Subject: [oslc-core] Provided guidance for adding relationship labels Sent by: [email protected] I wanted to call out some specification updates that was created for handling of relationship labels on URI relationship properties. Note the support for this is optional but wanted to make sure this was done in a uniform way across implementations. Let me know if there are any issues with this. http://open-services.net/bin/view/Main/CmSpecificationV2#Labels_for_Relationships Thanks, Steve Speicher | IBM Rational Software | (919) 254-0645 _______________________________________________ Oslc-Core mailing list [email protected] http://open-services.net/mailman/listinfo/oslc-core_open-services.net _______________________________________________ Oslc-Core mailing list [email protected] http://open-services.net/mailman/listinfo/oslc-core_open-services.net _______________________________________________ Oslc-Core mailing list [email protected] http://open-services.net/mailman/listinfo/oslc-core_open-services.net
