What's the best start of cell marker for Python and Julia scripts? about jupytext HOT 5 CLOSED

Sure! Yet,

• Jupytext has a very different practice of cells markers (few explicit markers, start+end) than Hydrogen (all cells are marked, start only). This is also discussed here. Hydrogen syntax is simpler to parse. Jupytext's cells are more difficult to parse, but arbitrary python files can be converted to notebooks, and converted back to the original python files without loss or additional markers.
• The definition of markdown cells in Hydrogen is pending (nteract/hydrogen#1296, nteract/hydrogen#1183)

For now I would be in favor of

• preserving the current 'plain python' format for Jupytext's python scripts, and simplify # + {} to # + for explicit code cells with no metadata
• implementing a 'hydrogen script' format (R, Julia, Python) in Jupytext when specifications for markdown cells (and cell metadata) are chosen. The Hydrogen format must remain easy to parse (simply split on cell markers), as an additional JavaScript parser will have to be implemented in Atom.

from jupytext.

maybe stick to hydrogen
-> one standard less and hydrogen is pretty established.

Would hydrogen blow up when adding metadata like this?

# %%name-of-cell {metadata}

from jupytext.

That would be fine. Once Hydrogen sees ${comment symbol} %% it treats that line as the start of a block.

Source:
https://github.com/nteract/hydrogen/blob/58ba87e1747366c46f9baca87dcdda27401f6a5a/lib/code-manager.js#L138

from jupytext.

I've found more python scripts with cells defined with #%%! VS code has a plugin that uses that convention. It's also used by Spyder. No sign of markdown cells yet, though.

from jupytext.

I'll work on an Hydrogen/VS code/Spyder compatible format at #59 (see there for possible specifications of the format). With this issue I'll just simplify # + {} into the lighter # + when possible.

from jupytext.