Original Tex is not an option, since PDF, today's font formats (and Unicode) and many algorithms didn't exist when Knuth decided no features would be added anymore. Maybe I should hack the source code of pdftex or luatex? (...)
you want to writ ea full blown macro package with advanced font support etc etc? well, it will take you a couple of man-years
I'll never use macros, whatever the task is :)
well, for most users a macro is just an abstraction doing something typographically meaningfull; in that sense it's not much different from a library where you use some function call to achieve the same
What I like in non-macro programming is that typing (in the programming sense, where elements can be used only in places where their type, or class, is allowed) sometimes helps compose a lot of complex things into really simple things. I believe the translation of a document structure made of programming language data into typographic data is easier (mostly easier to debug).
no matter what programming tools you use, the problem remains the same (…) what system you use or build depend on what you need to typeset (…)
I do think there are language features that can make organization of documents and use of techniques more powerfull. But I can only prove that (even to myself) showing something working.
If that's not possible today, maybe I could at least write a nice library of general purpose typesetting algorithms (…)
... general purpose might not be what most users want; eventually luatex will provide many typesetting library components, but in themselves they are rather useless, as it's the integration that makes it useable.
That's good enough for me. Which books and articles do those people who are going to write those libraries read? Does TeX book contain explanation of those algorithms? (Although I've seen that Context can do things TeX itself can't, maybe there are new important ideas around.) Maurício