Best Practices Wiki

Handling large XML

• Use a streaming event-based parser (like StaX for Java) with look-ahead capability to avoid reading the whole XML file into memory. The event-based parser only uses a small amount of memory to store the section of XML it is processing, moving quickly through the file to find what it is programmed to looked for. Look-ahead capability avoids multiple reads when a certain subelement is desired.

• Shred into a database. Many libraries exist to create a temporary or long-term relational persistent store of the XML data. Look-ups are then fast and efficient. Delete the relational tables when you are done processing the XML, if logs aren't needed.

Operational Use of HMIS XML

• Use web services with an Application Programming Interface (API)
  • Two main types of web service
    • REST
    • SOAP
  • An API is a cookbook (not software) for things like "add client record", "get client record"
    • There is no standard HMIS API, though non-standard ones are in use between systems. An unofficial provider/vendor neutral email listserv and forum for this topic is here.