CMIS proponent/contributor Laurence Hart (aka @piewords) picked up on the latest debate on whether or not JCR will die at the hands of CMIS. There are a number of interesting points from the blog over the article in CMS Wire – not the least of which is that degenerative disease is less harsh than zombification – I won’t take time to argue that disturbing point but I do want to reiterate a few points on why or rather when you should be concerned with CMIS.
Despite the obvious advantage of a language neutral spec to the alternative, there is no guarantee that CMIS will be any more successful than its predecessors. I still maintain that there is no competitive advantage, only parity in supporting CMIS. Therefore mainline vendors may be inclined to keep their noses in it but they will not drive adoption. Smaller or open source ECM vendors will promote CMIS because it removes differentiation, leveling the playing field to a narrow set of capabilities and forcing the decision to focus on price.
Someone must lead the charge and that must be the consumers of the service. Those consumers are applications made from someone other than the ECM vendors themselves. Mobility is a great driver here. Application developers, even when designing for a specific backend can benefit from the debates that were part of the specifications development on what is and is not essential to content transactions. The fact that the resulting application is made more flexible is really a bonus. Similarly the peer-to-peer marshalling of content between cloud application services also as a similar benefit on a different scale. The “I” for interoperability in this case is far more important than the “M”.
Bottom lne if you are preparing an RFP for content technology, CMIS is certainly worth mentioning but is not a key component to the decision criteria. At least not yet anyway. Architects and developers however should absolutely consider it even if there is no direct requirement as they can benefit from the design and modeling work whether they implement the entire interface or not.