[PRL] akward and unwieldy

[Johan Ovlinger]

In semantics-seminar discussions over the years, various grousings
have been voiced about module systems becoming awkward on non-toy
problems as linking statements have failed to scale, or arcane
incantations being required in order to get the type-checker to
believe that two imported types are the same...

As a concrete example, I seem to recall Matthew Flatt claiming to have
solution for the akwardness of manually declarring all the linking
instructions in his Units-for-C work. About a year ago, I think.

But this is all very informal and handwavy.  Does anyone have any
pointers to published critiques?