[Axiom-developer] Re: defintrf.spad compiler bug

[Stephen Wilson]

address@hidden writes:

> I've identified 592 potential places where this could occur.
> Obviously it does not happen everywhere. I'm reviewing each
> code sequence to try to find the nature of the failure.
> After that I'll look into the compiler.

Im looking at a possible solution, but I am waiting on a build.  I
cannot say at this point if it is workable. 

In short, Im experimenting with querying the environment in
compiler.boot:compWithMappingMode and trying to determine the proper
logic which can reliably mark an identifier as denoting a Record
field.  If this is workable, then I belive that within the same
function we could choose to emit these labels as quoted symbols.

In fact, I am first trying a more general approach first -- If the
identifier does not have a mode in the current environment, assume the
identifier is not bound and quote it.  I hope to refine this if I can
show the basic approach is valid.


Take care,
Steve