Carl Mäsak
Carl Mäsak
Created #548 to track the breakage, since it's outside of the scope of this issue.
A related macro `deeplyChanged` could detect changes not just to the variable itself (which is still useful and exactly what we need sometimes), but to nested accesses to the variable.
I wonder if #186 could be given a semantics such that this issue can just use it straight off?
Just driving by to say I think the `while` in OP should have been a `repeat while`, especially so since I'd intuitively expect `changed(X)` for any `X` to be `false`...
Much later thought: this feature might be trying to be "too magic" by implicitly deciding when something changed (and injecting code at that point). With variables, it's easy to detect...
Oh! I just realized that this can also give us a solution to #158 and #210 which is (a) better than the `inner` declaration discussed in the former issue. Those...
Needless to say, this will delay the evaluation of the macro. Given how actions fire in preorder, we will always encounter first the macro and then the context. (Although we...
Even #13, it seems to me, might fit into this scheme. I haven't thought it all through, but it seems to me that `amb` is a macro in the `statement`...
> There are a lot of details to this to work out. For example, it strikes me that the junction operators have _context precedence_; that is, `&` binds tighter than...
One implicit but unaddressed thing in this issue is that the macro is having an "out-of-body experience" in the sense of being called at the point of its context-providing ancestor....