Compiling Applications

NeuroKernel Applications are compiled using a SDK which contains the API for the system. Initially, the develpoment SDK is Java based. We will provide support for other programming languages that compile to either Java ByteCode, Javascript or WebAssembly. In the future, NeuroKernel dynamic edition may also ported to other server side platforms other than Java servlets.

1. Writing Portable Code

NeuroKernel API tries not to use an extensive set of Java packages for portability. For an API, less libraries and methods you use for the target platform, more portable your code will be. Developers are free to write complex applications with deep dependencies. NeuroKernel API is also Java 6 compatible which means legacy platforms can also be supported. If a developer wants their NeuroKernel code to be portable, use only NeuroKernel supplied API and check if a package or method for the transpiler used is available on other supported platforms as well.

2. Compiling to ByteCode

Compiling a NeuroKernel application to Java byte code is the simplest approach in terms of compiling the code. Adding NeuroKernel API jar file to the classpath is all is needed.

2.1 Servlet Container

Developers can use any servlet container that implements Servlet 3.0+ specification. If you have a legacy server with a older Servlet specification, please contact us. We can provide you with a legacy build; however, additional license fees may apply. It may be required to put the servlet API and other related libraries on the classpath when compiling.

2.2 Command Shell

In Dynamic Edition, it is possible to open a VT terminal to your server where servlet container runs, and compile and run applications from the command line. It is even possible to plug command line applications into other applications. The display port is automatically set by the VT terminal and read by the compiled application. There is another library API file to use while compiling command line NeuroKernel applications. Terminal IO only works with NeuroKernel terminal, and it wont work with command line applications. If there is demand, we may offer an emulation for command line console input output which would function exactly same as NeuroKernel terminal. Command line applications are an important tool for developers to rapidly test or debug their code.

3. Compiling to JavaScript

Compiling your NeuroKernel application code to JavaScript is very straight forward if you have experience with the transpiler you use. The size of code produced depends on the transpiler. NeuroKernel API has annotations for code generation which may help reduce the size of the code by isolating unused features. Transpilers tend to remove unused classes and methods as well.

4. Compiling to WebAssembly

NeuroKernel WebAssembly support is still in the works, but we are pretty close. NeuroKernel has a high performance protocol, and that is going to be a winner for WebAssembly applications that does not want to use only Canvas to render. Please see our Experiments page for the details of the current WebAssembly support.

5. Using Generators

When an application is compiled from Java to Javascript or WebAssembly, it may need some generators to execute at client side such as embedding resources in the executable Javascript file. If compiled application will be available as remote application, a bootstrap html file is also needed to be created. NeuroKernel generators will help creating the required files along with the compiled executable. A set of annotations are supplied with the API to configure the generators that may be used on NApplication class when extended. Some transpilers supply tools to make generating these files easier, for others API has annotation processors available to be used under generator package.

5.1 Annotations

NeuroKernel API has annotations that control the code generation while compiling the application. These annotations are used either by a compiler based API or an annotation processor that is available from the NeuroKernel SDK.

5.1.1 Configure Annotation

Configure annotation is the main configuration annotation for NeuroKernel application. It can also be used to optimize the size of compiled code.

5.1.2 Executable Annotation

Executable annotation is used to generate application launch code. An application bootstrap code can be written by the developer to support a specific launching process which may be passed as an argument to the generators.

5.1.3 Remote Annotation

Remote annotation is used to generate code for applications targeted for remote access. It can be used also to configure Executable annotation. The “rexec” is the command that can be used to execute a remote application via the IRuntime interface.

5.1.4 ApplicationCache Annotation

NeuroKernel client side applications can be cached to the browser for offline access. This may be useful if the site to the application is down. the Service workers are used for this including a fallback to AppCache mode. There may be client based limitations that could be solved in a later version. Please take note that, the deployment must have HTTPS certification for this to work.

6. Supported Transpilers

NeuroKernel Application can be transpiled to JavaScript using a supported tool. As of this version, we fully support TeaVM, GWT and J2CL transpilers. Others transpilers for Java or other languages will be made available. Experimental builds can be accessed from Experimental Product Builds page. We also plan to fully support WebAssemly in the next release.