Scheme eval issues

Jens Axel Soegaard <invalid@soegaard.net> writes:

>  === context ===
> /Applications/PLT Scheme v3.99.0.12/collects/scheme/private/misc.ss:63:7

That is .  Line.. AND column numbers?  File.line.column can
reference unnamed and ambiguously named procedure text unambiguously!

How does that work with macros?  I heard a rumor on the street (Scheme
St.) that someone did a thesis (long ago?) about mapping from machine
state back to source code (file.line.col?) THROUGH (simple?) macro
transforms.  If that thesis exists, is it available on the web?

Matt Birkholz skrev:
> Jens Axel Soegaard <invalid@soegaard.net> writes:
> 
>
> That is .  Line.. AND column numbers?  File.line.column can
> reference unnamed and ambiguously named procedure text unambiguously!
>
> How does that work with macros?

Could you elaborate the question?

--
Jens Axel Søgaard

Jens Axel Soegaard <invalid@soegaard.net> writes:

> Matt Birkholz skrev: 
>
> Could you elaborate the question?

If I evaluate this special form

(dotimes (i 'a) (display i) (newline))

and it expands to

(let ((bounds 'a))
(let loop ((i 0))
(if (fix:< i bounds)
(begin
(display i) (newline)
(loop (fix:1+ i))))))

I would expect a runtime error message like

;The object a, passed as the second argument to fix:<, is not the
;correct type.

but what source code location would be referenced, and how -- under
what conditions?

A 90% solution that was published 15 years ago would be a patent-
defensible boon.  It would be  if some of the Scheme work from 20
years ago gradually appeared on The Net (i.e. with few copy rights
withheld), so that it can be easily found and referenced as prior art.