plan98 VS rapidyaml

thank you!

if it helps, my last commit touches the primary aspects of modules.

overall, i’m integrating a test payment flow on https://thelanding.page

this commit[0] i export two functions from my payment debugger module (i am intentionally overloading the word module to additionally bind a view; this is a personal hot take and not general guidance, just in case you see how i use ‘@sillonious/module’).

so i expose payment-debugger.js in the index.html file import map as @sillonious/payments.

i import that module in the message-pay component.

when the user clicks buy now, i call the newPayment function (which ultimately lands in client.js, after traversing the network; in case you’re curious about the full route). i also set a timer that’ll check the payment status every five seconds for 15 minutes. displaying a timeout message is a not yet implemented edge case.

[0]: https://github.com/tylerchilds/plan98/commit/db3c037345077fe...

Interesting, I hadn't come across extism before. How hard would it be to package https://github.com/biojppm/rapidyaml in this way? (And do you have a extism for dummies guide?)

The inability, or brittleness, to embed NUL bytes into the string, for once. Zeroed bytes can be valid as an internal bytes of a longer encoded character. And the inefficiency of tempting every caller to rederive the string length on every use, leading to such bugs as quadratic parsing behavior with sscanf. The extra register for an explicit length is a very minute price to pay compared to that.

Although you still have the other problem of sscanf, where read times can become quadratic.

You are really simplifying that issue. A lot of it has to be how the implementation of sscanf works and it's super easy to abuse by complete accident when a lot of data needs to be read and parsed to correct values. See more here: https://github.com/biojppm/rapidyaml/issues/40

Leading to eye-rolling problems like these: https://github.com/biojppm/rapidyaml/issues/40