The history of the relationship between coffins and page-makeup is relevant here.
Briefly, the ideas behind the current implementation of coffins was originally related purely to typesetting, in particular to the complex alignments that are needed in maths typesetting of, for example, three matrices that have to align row 3 of one with the middle of the next and the last row of the final one.
The next development stage was somewhat different, aimed at page-makeup of title pages etc without using absolute coordinates or any “draw-like techniques”.
I shall leave out my comments on whether it is sensible to implement the construction of such pages without using draw-like technology.
But LaTeX definitely needs a “page-construction mode” that (amongst other useful stuff) can be used to construct such single (and singular) pages. This will combine something based on the draw technology with other layout technology including (I hope) an extension of the coffin technology.
This extension will, for example, support integration with draw technology and also support more complex alignments of nested coffins, as needed by the maths examples above. But as you remark, come the day!
So now is the time to provide good specs for the “New Coffins from Old” project :#)
I'm mindful of the link, but at present can't say what exactly the
result will be :)
I suspect this will need to be explored once a more fully worked-out
l3draw is finished. Possibly coffins will be used instead of nodes ...
For l3draw, it does make a kind of sense to hand-off typesetting actual text to the coffins module.
For absolute positioning, I wonder whether it might be reasonable to have implicit coffins (with nothing inside them, but with poles other coffins can be attached to) for the page and/or text-block.
Then again, creating these might be functions of the eventual L3 successor to the geometry package. (Headers and footers and margin notes are all going to be in coffins, come the day—right?)