James Snell has just published an article on developWorks, which illustrates how to use the Atom Publishing Protocol to publish Common Alert Protocol (CPA) alerts. CAP defines a XML data model for specifying hazardous alerts and notifications. The article uses the Apache Abdera implementation of APP to indicate how to publish, modify and delete CAP alert documents.
Thanks Jim. It is a good article but says little about the specifications that typically come with the requirement to apply CAP. IOW, a police incident is not a health incident is not an outage incident and so on. The problems of mapping event types (in the generic sense of the term not the local CAP sense) to alert types are daunting. Where CAP is required, the specs usually also specify ebXML/WebServices not RSS/Atom. Also, since an XML consumer/producer is a fairly easy thing to implement these days, crushing into the RSS/Atom formats doesn't buy me much.
I don't disagree that this can be done. I'm just not sure the author is looking at the other 95% of the technical requirements for alert systems where CAP is about 5%. That said, a dashboard for Alerts that also includes an RSS Reader is a much more capable dashboard.
len, the idea with this article was just to present the basic publishing mechanism. I'm fully aware of the other aspects of the problem but there simply is not enough time and space to cover all of it in a single developerWorks article.
The basic publishing mechanism is well-known. Without other aspects of the problem, it becomes an adverstisement for IBM solutions to the HLS initiatives.
1. For example, CAP implementations often require a means to postback to the original source if attachments are added by direct or cascading reports. Does a CAP implementation help or hinder that process?
2. Are RSS/CAP implementations 'reliable' in the sense of creating a 'reliable messaging system' or only as reliable as the basic xmlhttp infrastructure?
3. If one has an xmlhttp infrastructure plus XSLT, is there any reason to prefer the RSS/Atom formats given CAP is already in an XML format?
There is considerable promise in the alert systems as a piece of the overall HLS infrastructure. RSS/Atom may have a role to play here but only if the overall requirements can be met. Otherwise, it is another case of a web supporter base hijacking a national initiative for its own growth without regard to the health of the host.
So please do follow up with more articles. This one is a good start. I am challenging the premise that it is a good way to do CAP work not because it can't be part of a solution, but because any limits have to be understood ahead of time. Otherwise, procurements stall while developers try to paper over the gaps.