custom, persistent variables for code snippets

[thartbm]

Somehow / somewhere, each code snippet can specify custom variables to be kept in memory, and to be stored.

Maybe a json, with the same base filename as the code snippet (assume none, if there is no matching file).

Which has entries for each variable, specifying:

• variable name
• variable type (numeric, boolean or character?)
• anything else?

These variables are added as columns / dict entries to the performance matrix. That is. they will be available to the code snippets, and they will also be stored in the per-trial files.

[thartbm]

Hmmm. Let's think this through.

1. The trialstate (both the transient and persistent part) should be readable to code snippets.
2. In case code snippets / scripts need to store stuff for future reference, that should happen in the persistent part. Perhaps something called user / script / custom / something else, which can have whatever?
3. Then finally... some of this might also end up in the performance matrix, but that needs to be specified somewhere. In the script .py file would be convenient for users, but hard for the experiment, so let's think about that a bit better. Alternative are: 1) in the scripts folder in a .json that has the same filename as the script (except for the extension), this would require a little extra work, but things are kept together, or 2) in the main experiment json, which could require less work, but does mean that stuff is not together.

The first two parts can already be implemented.