What's new?

In 0.50.2

• Many bug and stability fixes
• ReturnFullResource will not only set the Prefer header, but will do a subsequent read if the server ignores the Prefer header.
• Client will accept 4xx and 5xx responses when the server does not return an OperationOutcome
• Client gives clearer errors when the server returns HTML instead of xml/json
• Call signatures for OnBeforeRequest and OnAfterResponse have been changed to give low-level access to body and native .NET objects. OnAfterResponse will now be called even if request failed or if response has parsing errors.
• The FhirClient has a full set of new LastXXX properties which return the last received status/resource/body.
• Serializers now correctly serialize the contents of a Bundle, even if summary=true

In 0.20.2

• FhirClient updated to handle conditional create/read/update, Preference header
• Introduction of TransactionBuilder class to easily compose Bundles containing transactions
• Model classes updated to the latest DSTU2 changes
• Serialization of extensions back to "DSTU1" style (as agreed in San Antonio)

In 0.20.1

• Added support for async

In 0.20.0

• This is the new DSTU2 release
• Supports the new DSTU2 resources and DSTU2 serialization
• Uses the new DSTU2 class hierarchy with Base, Resource, DomainResource and Bundle
• Further alignment between the Java RM and HAPI
• Support for using the DSTU2 Operation framework
• Many API improvements, including:
  • deep compare (IsExactly) and deep copy (DeepCopy)
  • Collections will be created on-demand, so you can just do patient.Name.Add() without having to set patient.Name to a collection first
• Note: support for .NET 4.0 has been dropped, we support .NET 4.5 and PCL 4.5

In 0.11.1

• Project now contains two assemblies: a "lightweight" core assembly (available across all platforms) and an additional library with profile and validation support.
• Added an XmlNs class with constants for all relevant xml namespaces used in FHIR
• Added JsonXPathNavigator to execute XPath statements over a FHIR-Json based document
• Added a new Hl7.Fhir.Specification.Source namespace that contains an ArtifactResolver class to obtain schema files, profiles and valuesets by uri or id. This class will read the provided validation.zip for the core artifacts. For more info see here.
• Changed FhirUri to use string internally, rather than the Uri class to guarantee round-trips and avoid url normalization issues
• All Resources and datatypes now support deep-copying using the DeepCopy() and CopyTo() methods.
• FhirClient supports OnBeforeRequest and OnAfterRequest hooks to enable the developer to plug in authentication.
• All primitives support IsValidValue() to check input against the constraints for FHIR primitives
• Models are up-to-date with FHIR 0.82
• And of course we fixed numerous bugs brought forward by the community

In 0.10.0

• There's a new FhirParser.ParseQueryFromUriParameters() function to parse URL parameters into a FHIR Query resource
• The Model classes now implements INotifyPropertyChanged
• FhirSerializer supports writing just the summary view of resources
• Model elements of type ResourceReference now have an additional ReferencesAttribute (metadata) that indicates the resource names a reference can point to
• ModelInfo now has information telling you which FHIR primitive types map to which .NET Model types (this only used to work for complex datatypes and resources before)
• We now support both .NET 4.0, .NET 4.5 and Portable Class Libraries 4.5
• For .NET 4.5, the FhirClient supports methods with the async signature
• All assemblies now have their associated xml documentation files bundled in the NuGet package
• Models are up-to-date with FHIR 0.80, DSTU build 2408

In 0.9.5

This release brings the .NET FHIR library up-to-date with the FHIR DSTU (0.8) version. Additionally, some major changes have been carried out:

• There is now some documentation
• The FhirClient calls have been changed after feedback of the early users. The most important changes are:
  • the Read() call now accepts relative and absolute uri's as a parameter, so you can now do, say, a Read(obs.subject.Reference). This means however that the old calling syntax like Read("4") cannot be used anymore, you need to pass at least a correct relative path like Read("Patient/4").
  • Since the FHIR create and update operations don't return a body anymore, by default the return value of Create() and Update() will be an empty ResourceEntry. If you specify the refresh parameter however, the FHIR client will immediately issue a read, to get the latest updated version from the server.
  • The Search() signature has been simplified. You can now either use a very basic syntax (like Search(new string[] {"name=john"})), or switch to using the Query resource, which Search() now accepts as a (single) parameter as well.
• The validator has been renamed to FhirValidator and now behaves like the standard .NET validators: it validates one level deep only. To validate an object and it's children (e.g. a Bundle and all its entries and all its nested components and contained resources), specify the new recursive parameter.
  • The validator will now validate the XHtml according to the restricted FHIR schema, so active content is disallowed.
• The library now incorporates the 0.8 version of the Resources. This means that developers using the API's source distribution need only to compile the project to have all necessary parts, there is no longer a dependency on the Model assembly compiled as part of publication. Note too that the distribution contains the 0.8 resources only (so, no more Appointment resources, etc.).
• The library no longer uses the .NET portable class libraries and is based on the normal .NET 4.0 profile. The portable class libraries proved still too unfinished to use comfortably. We've fallen back on conditional compiles for Windows Phone support. Cross-platform compilation has not been rigorously tested.
• After being updated continuously over the past two years, the FHIR client needed a big refactoring. The code should be readable again.

Before

Is history. If you really want, you can read the SVN and Git logs.