2019-06-12, 08:16 PM
I ran something similar on my machine and it takes roughly 1 ms for the first parsing+rendering then about 100 µs for each subsequent parsing+rendering. If you're looking for sub-millisecond processing, the library is designed to separate parsing from rendering in order to persist a pre-parsed representation of the text to the database and only render it dynamically. It's a trade-off that makes it possible to render a text in a tenth of the time it would take to parse while being able to use dynamic templates.