On Sat, Aug 29, 2009 at 1:10 PM, Aubrey Jaffer<[email protected]> wrote:
>
> The number of SRFIs supported by the current release would be a good
> metric of an implementation.

Yes, but it seems that the SRFIs themselves should be weighted in some
way.  For instance, SRFI 6 (Basic String Ports) is nearly universal, so an
implementation that omitted SRFI 6 should be penalized.  But SRFI 21 is
only supported by Gambit.

> The list is out-of-date.  SCM also supports 96 and 98.  SLIB also
> supports 23 and 96.

I'll be updating it as I get more information.

> Any implementation supporting SLIB (Bigloo, Chez, ELK 3.0, Gambit 4.0,
> Guile, JScheme, MacScheme, MITScheme, Pocket Scheme, RScheme,
> Scheme48, SCM, scsh, SISC, Stk, and VSCM) supports these SRFIs through
> SLIB: 0,1,2,8,9,11,23,47,59,60,61,63,94,95,96

I guess a second column would be necessary, then.

>  | Not only did I have the temerity to measure Scheme implementations
>  | on a hidden scale (it could have been a Ouija board), I had the
>  | temerity to *rank* the implementations.  Furthemore, I divided them
>  | into four broad tiers.
>
> Your classification adds no objective information to the table,

Admittedly it doesn't add much, but there is no point in pretending that
all implementations are equal.  PLT Scheme is clearly a more important
implementation than VSCM.

> while
> it ignores the most important distinction between implementations:
> whether they support R4/5RS or R6RS.  Programs are not interoperable
> between R6RS and R4/5RS, so that is the most important distinction to
> make in your table.

I have deliberately omitted that distinction.  I will explain why later.

> Please remove your arbitrary tiers and classify implementations by
> which standard(s) they hew to.

No.  But you should feel free to make your own spreadsheet or document.

-- 
~jrm

_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss

Reply via email to