There are some situations where character recognition is easy enough that it should be possible without a ton of machine learning, so I am creating a JS/wasm client-side library that can accurately identify characters from pen or touch strokes. For some reasonably accurate results for the English alphabet, it currently zips down to about 50KB all-inclusive and takes just a few ms to identify a character on most devices.
I have tried to make it easy to train your own character set if anyone wishes to try that out. It is still in the early stages but I am hopeful it will eventually be good enough for many use cases.
There are a couple good libraries to display LaTeX, but there aren't enough ways to generate LaTeX without a human who knows the syntax. So I'm trying to make it as simple as possible to generate LaTeX for at least basic math expressions.
Also, I want the contribution process to be a bit different. It's a work-in-progress but it should eventually be very simple to make changes, test them, and submit.
Mathzetta is still early stages and any feedback is appreciated.
You can check that random checkbox in the Content button to intercept tabs. I know it is not obvious.
I really don't know whether iframes should be enabled by default. It was the default, but I kept clicking endnote links and decided that no iframe is a better default.
Thanks! It is intimidating to build from the ground-up, but it has been worth it.
I'm not exactly sure how I want to integrate this into a more all-in-one solution. Once I check off some more immediate items on my to-do list I will try to figure out the best way to do that.
Thanks for the kind words! Mainly, I don't really want to deal with pull requests at the moment. Also, I want a dogfooded readme and so having an additional Github readme would be a bit of trouble.
I wanted to create javascript-free documentation and articles, but it was annoying to add latex, code, sidenotes, a table of contents, and other niceties. So I created a tool to automate the process. It is very unfinished and can't handle many edge cases, but there are plenty of features to try out.
Learn how to use it from the documentation located at https://triplelog.com/writer/readme.html. To give you an idea of what is possible, that page was created with Triplelog Writer and is entirely javascript-free. The total network load is about 100 KB - most of which is the Katex CSS and fonts.
I think it is really easy to use, but the documentation is probably not good enough yet to make everyone agree with that statement. If you have any questions about how to use it, ask away.
I have tried to make it easy to train your own character set if anyone wishes to try that out. It is still in the early stages but I am hopeful it will eventually be good enough for many use cases.