A RxTender backend is a python package that contains jinja2 templates. A backend can implement one, several, or all features of RxTender : Framing, serialization, and router. Each of these features is implemented as several template files. The following figure shows how code is generated:


RxTender takes an rxt definition file as input. It then generates chunks of code from each template file being used. Finally, all chunks of code are combined to generate a final file.

Backend Structure

python package, jinja2 template engine for code generation.

The source layout of a backend is the following one:


where foo is the name of the backend, [language] is the name of a programming language. There is one [language] module per programming language supported by the backend. An optional "framework" name can be provided if different frameworks of a language are supported and they must generate different code.

All templates follow the same naming convention:

[output type].[template type].tpl

[output type] is the type of file to generate. All backends implement the "source" output type. The "source" templates generate source code. If the selected language also needs other files types to be generated, then additional output type templates can be implemented. For example, with C or C++ a header file is also needed. So, these languages also implement "header" output types. So a backend with support for c contains "source.content.tpl" and "header.content.tpl" templates in the serialization modules.


  • stream creation request
  • stream creation ack
  • stream creation nack
  • stream deletion request
  • stream deletion ack
  • next item publish
  • stream complete
  • stream error


  • header.tpl
  • content.tpl
  • footer.tpl


Streams generate Observables proxies that follow the Observable Contract