So now infix operators work, sort of; there's still no precedence rule support in the new parser, which will be a righteous pain in the backside to reimplement. Instead of tackling that straight away, I've decided to go back to fundamentals and figure out why functions are no longer capable of returning anything. (In case you hadn't guessed, in a language that aspires to support pure functional programming as richly as possible, return values are kinda important.)
In the process I decided to sacrifice one of the oldest sacred cows in the Epoch toolbox, which is multiple return values. Epoch's first incarnation allowed functions to return multiple values via a tuple mechanism; unfortunately, tuples and structures became extremely similar, to the point where the distinction between them was largely academic and historical and not really practical at all. Since pragmatism is the guiding rule of the Epoch project, I figured it makes the most sense to axe tuples and collapse everything of that nature into structures.
Which brings us back to multiple return values: there's no point in having them anymore, if one can just return a structure instead. I may end up implementing some syntactic sugar support for anonymous return structures if it ends up being useful (God knows I love Lua's multiple-return system... well, sometimes). It shouldn't be too hard to add sugar to the new parser structure, because the goal of the new parser is basically to make everything syntactic sugar.
Anyways... I could ramble for a bit on how the new architecture will basically turn Epoch into a programming language generator rather than a strict language itself, but I should probably organize those thoughts a bit more before splattering them all over the page. So more on that later.
For now, back to making functions useful again!