Procedures

One of the strengths of Beaker is that it can execute code in the subkernel, and use the results/output of that code in Beaker itself.

This is accomplished by use of “procedures”, snippets of code that are executed in the subkernel environment, as if it were run in a notebook cell.

Procedures exist at the context level, and are tied to that particular context. A single procedure can be used for multiple subkernels if the context is designed to handle them.

Each procedure is defined as a file that looks like a normal code file, but is actually a Jinja template, which is rendered at time of use to generate the code that is to be executed. This allows one to provide “virtual arguments” in the form of template substitutions or rendering prior to the code being run.

Procedure definition

The directory which holds the context class definition file should also contain a directory named procedures. In this directory should be one or more sub-directories whose names match a subkernel slug name. Your procedure template files should be placed inside this subkernel folder.

A procedure’s name should match the filename, minus any extension. For convenience, the extension of the language used is often used. I.e. .py for Python files, .jl for Julia, etc.

If a context is to work with more than one subkernel, procedures executing equivalent code should exist for each subkernel language. In this case, the file names, excluding extensions, must be the same.