I am considering using NHapi in some HL7 related projects. Usually when I decide to use any open source library in a project, I have two criteria:
Looking at the NHapi forum on SourceForge, it does not seem to fulfill any of the above two criteria.
The other options are either to buy a commercial product or to write a parser.
Anyone has any suggestion or thoughts of using NHapi ?
On
We evaluated NHAPI and decided not to use it for the same concerns you cited. Instead we went with HL7 Spy. It has a handy GUI client for sending messages (useful for testing) as well as a DLL that helps you build up messages.
Unfortunately, as you mentioned, it's a commercial product and not open source. But we've been pretty happy with it.
On
We decided to use it in an integration engine. My impressions:
We found that the API object model is confusing and not homogeneous when using the different HL7 versions (V231 and V230).
We've also found some bugs when parsing text messages.
IMHO NHAPI is not unreliable, but before using it, evaluate the API trying to test all use cases in which you need NHAPI.
After all the experience with NHAPI, what I can say 100% sure is that if we have had time, we would have developed our own HL7 API.
Hope this helps.
We have begun implementing NHAPI in some of our HL7 processing applications. We had the same concerns as you did, but given that it is open source, it is certainly more useful than writing your own parser. Since it and the HAPI project on which it is based are licensed under MPL, you could always fork the codebase if you find the project is not meeting your needs.
We have also used a commercial product whose name I forget, but that has caused its own challenges. The installation and licensing is a challenge, especially on newer operating systems and the company is de-emphasizing the product, so support is very poor.
I have found also that there's at least a little bit of third-party usage out there too: http://dib0.nl/code/255-where-to-begin-if-you-want-to-start-with-hl7-in-c-or-java