Re: Should we have (srfi srfi-N) modules for all SRFIs?

[Rob Browning]

Neil Jerram <address@hidden> writes:

> Hmmm...  But only if people include (use-modules (srfi srfi-N)) in
> their code.

In the long run, I think that the srfi stuff should proabably be
integrated somehow with SRFI-0 and/or SRFI-7, allowing people to write
more portable code:

  (cond-expand
    (guile (use-modules (srfi srfi-1)))
    (rscheme ...)
    (else))

  ...

  (cond-expand
    (srfi-1 (define my-filter filter))
    (else (define my-filter local-filter-implementation)))

etc.

> If we did this, I'd prefer to see a corresponding extension of the
> deprecation system as well such that use of a srfi-N procedure
> _without_ the appropriate `use-modules' would issue a warning.

Perhaps not a bad idea, though only appropriate for deprecated access
methods since there may be multiple acceptable ways to get the same
functionality (i.e. module aliases or alternate locations).

-- 
Rob Browning <address@hidden> PGP=E80E0D04F521A094 532B97F5D64E3930