touying better animations and html support

[zral0kh]

HI i just came across this interesting package and i am quite impressed.

I am one of the recent active maintainers of touying and have thought of providing more finegrained control similar to manim e.g. via stacked transformations as you do already. (thus i introduced the waypoints bc i did hate counting subslides since ever)
I will look into this package to understand how it works and potentially adapt some of it.

The reason i am writing this is that i do now also think that with html export maturing we could actually bring such things to the browser, with math being mathml and asset bundling, etc, exciting stuff.

i understand that you want to work on a clean slate and not have to struggle with touying' s legacy code and functionality, but if you would be open, i'd love to share ideas.

[pacaunt]

The reason I branched out for this separate package is not because of the Touying's legacy code compatibility, it is the concept of touying core implementation.

I have a strong believe that content is a type of data that should be inspected only when necessary, not the expected functionality, and cannot be totally reconstructed. Touying uses content parsing and walking down the data tree to obtain the information and transform the content, which lead to two limitations:

1. It cannot inspect layout and context. However, a workaround can be done by wrapping the inner content in the form that Touying understands.
2. It cannot parsing deep composite functions when they are not returning content. A simple example is the group or other grouping functions of CeTZ cannot be inspected because they return an array of functions. Touying can overcome this by using pure slide's method (uncover, only functions). However, with this approach, the slide engine does not know the existence of those call, therefore it is fundamental limitation that Touying cannot implement pause inside it.

That is actually the reasons why presentate and sanor exist, as my core idea to make presentation that is always compatible with any data types: you can do animations even with array, deep composite functions, context, string, integers. However, it does have a caveat: verbose APIs are needed.

In the future, may be very long future, Typst team plans to remove the content type and let users create their own types. I think this can totally break Touying, because, to reconstruct contents, you need to specify which fields are positional or named. Of course, you cannot know which the fields from users definition for sure, so there is no way we can parse and reconstruct the content as we do now.

You can grab my idea and implement it in Touying, of course. It is a good functionality for people there. But I just want to let you know the reason for separating the package, although my skill is really not enough to support the html and other fancy toolings other than writing a pdf.