Re: [PATCH] Source properties on arbitrary non-immediate values

[Ludovic Courtès]

Hi,

Neil Jerram <address@hidden> writes:

> That's fair enough.  I guess the rationale is that the unit of
> evaluation (as presented in backtraces for example) is a list, so it
> is useful for source properties to be stored on lists when those are
> read.

Sure.

> Yes, but why is that useful?

Why is it useless?  ;-)

I found it useful in a project that evaluates source in several steps:

  read [sexp] -> convert to alternate representation -> write things

Errors may occur during the last stage.  However, the user doesn't care
about the intermediate stage: they just want to know how the errors
occurring in the last stage relate to its source.  Therefore, source
information needs to be "piggy-backed" all along the process.

In any case, it's up to the user to decide what's useful and what's not.
Guile is here to implement mechanisms, not policy.  If we were to choose
the status-quo, then I'd have to implement a very similar mechanism by
myself, just because the one provided by Guile is unnecessarily
over-specific.

> (So far, I think I'd vote for fixing the manual rather than extending
> source properties ...)

I consider the restriction to pairs arbitrary (but I do understand that
it doesn't harm given the way it is currently used).  What's wrong with
removing such arbitrary restrictions?

Thanks,
Ludovic.