On Thursday, June 2, 2016 at 10:06:25 AM UTC-7, Matthew Butterick wrote: > I have a solution to this problem — is it legit, or is there some slick > Rackety technique I'm missing? > > I'm making a toy #lang interpreter for Basic, which allows variables to be > created with an assignment statement (like Python). The wrinkle in Basic is > that the execution order of the lines isn't known till runtime, so there's no > way to identify the "first" use of a variable and change it from a `set!` to > a `define` (e.g., by testing `identifier-binding`) > > To fix this, I gather a list of variable identifiers from the parsed AST, > insert the `define`s at the top of the `#%module-begin`, and then the rest of > the AST, and the expander proceeds. That works. > > But it seems like the gathering of variables should be able to happen as the > expander encounters them in the code (ie., do everything with one pass rather > than two). The problem is that by the time all the variables have been > discovered & gathered, the expansion is done, so AFAICT I can't go back and > `define` them retroactively.
I think you can do this by having your language provide it's own #%top syntax that defines what to do with unknown variables. http://docs.racket-lang.org/reference/__top.html?q=%24%25top#%28form._%28%28quote._~23~25kernel%29._~23~25top%29%29 -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
