John/Andy, I am not a fan of multi-create since I don't see what's wrong with the client just doing multiple POSTs.
However, if OSLC wants to support multi-create, then perhaps the way to do this is using the multi-part MIME type. The request would contain one part for each creation request, the response would contain corresponding parts for the results. Another route to multi-create might be SPARQL Update. Regards, ___________________________________________________________________________ Arthur Ryman DE, PPM & Reporting Chief Architect IBM Software, Rational Toronto Lab | +1-905-413-3077 (office) | +1-416-939-5063 (mobile) From: John Arwe <[email protected]> To: [email protected] Date: 10/06/2011 12:59 AM Subject: Re: [oslc-core] Creating multiple resources in one POST? Sent by: [email protected] > Can this [batched creates] be considered as part of OSLC core? "In theory", anything could be. I think the pragmatic issues have to do with consensus. The idea of posting to a collection when you want the create semantic is not something OSLC invented at all. It was in fairly widespread use in the wild already; AtomPub is the most commonly cited "original spec source" for the pattern, but I suspect lots of people would recognize it as factory create. OSLC Core "just" took that pretty well accepted Best Practice, applied it to RDF data, and *wrote down* the result -- the latter being a gap that is being recognized in the Semantic Web community generally. Some invention along the way was necessary (RDF/JSON) since the larger community had not prioritized that. In practice, I think it's a harder row to hoe with batched-anything. There is a much higher proportion of "invent new" relative to the case above, and no existing community consensus I'm aware of to "just write down". Implementations are headlights for spec work, have to be careful not to over-drive them. > And then it really helps the users to be able to create multiple resources with a single POST. This seems like it forgets about the error handling case you already acknowledged. Life is [relatively, still have to solve Location] easy when it all works. It's the partial error handling where it becomes difficult to achieve consensus - if you require ACID behavior, that's a fairly stringent server implementation requirement. If you don't, well-behaved clients may have a lot of work to do. Don't get me wrong, batched requests generally show up in many of the scenarios I'm expecting to see flowing soon from Tivoli; updates + creates. Maybe that eventually leads to a pattern people agree on. It simply does not appear to be imminent based on the actual scenarios coming in so far. Best Regards, John Voice US 845-435-9470 BluePages Tivoli OSLC Lead - Show me the Scenario _______________________________________________ Oslc-Core mailing list [email protected] http://open-services.net/mailman/listinfo/oslc-core_open-services.net
