Is this better merged into T298633 as a general graphical-programming-language support task?

I'd like to be involved in this investigation.

See, for example:

• https://en.wikipedia.org/wiki/File:Wikimania_2019_-_Multilingual_JavaScript.pdf
• https://en.wikipedia.org/w/index.php?title=File:Wikimania_2019_-_Let%27s_Completely_Change_How_Templates_Work!.pdf&page=30
• https://en.wikipedia.org/w/index.php?title=File:Wikimania_2019_-_Let%27s_Completely_Change_How_Templates_Work!.pdf&page=31
• https://github.com/cscott/turtlescript

Scratch and eToys are good models to use, FWIW, but I've got a personal lean toward things with 1:1 mappings w/ (a subset of) JavaScript (like the turtlescript example above). Note that I'd almost certainly use Blockly as the UX layer for turtlescript if I were building it in 2021, instead of hand-rolling my own UX.

(a perhaps non-obvious anecdote is that localization is/was a big issue w/ block-based representations. some languages have wildly varying string lengths to describe certain programming concepts, and some of these stretch the graphical layout of a 'block'. eToys and Scratch made it work, but the localization effort was more involved than "merely" translation.)

I wrote an experimental script to convert blocks from the Visual Programming languages Snap! and Scratch into source code based on a Mapping in a CSV-File. You can find the source code here. It is experimental and I am interested and try to extend the functionality of it. For example adding support for nested functions and also a reverse block create script. So you have source code as Input and get Blocks in a visual programming language as output.