[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Semantics of IMPORT-IMPLEMENTATION



>>>>> "Donovan" == Donovan Kolbly <donovan@xxxxxxxxxxx> writes:

Donovan> The specification part of the suggested revision to SRFI-0 does not
Donovan> seem to indicate the meaning of the multiple feature identifiers which may
Donovan> be present in IMPORT-IMPLEMENTATION.

Donovan> I assume that the bindings associated with all the named features are to
Donovan> be imported...? 

Yes.  That was a syntactic oversight I have hopefully corrected.  It's 
surprisingly awkward (for me, anyhow) to formulate this right.

Donovan> Also, I can't tell if it satisfies the specification for a system to do
Donovan> the work IMPORT-READER-SYNTAX via IMPORT-IMPLEMENTATION.  That is, could a
Donovan> SRFI-0 (revised) conforming implementation have IMPORT-READER-SYNTAX be an
Donovan> alias for IMPORT-IMPLEMENTATION? 

It is certainly not the intention because of the semantic ambiguities
associated with this.  What does it mean for IMPORT-READER-SYNTAX (or
an IMPORT-IMPLEMENTATION performing its job) to appear anywhere but at
the beginning of a syntactic processing unit?  Does it retroactively
affect the preceding part of the processing unit?  What is its scope
if it occurs inside a toplevel BEGIN?  And so on.

I actually do think the proposal makes these issues clear, but I'd be
glad to hear suggestions for clarifying this point.

Donovan> (I am assuming the general rule that, unless otherwise specified, a system
Donovan> can accept erroneous input.  That is, it may be an error to put
Donovan> IMPORT-READER-SYNTAX elsewhere in the file, but I'm not obliged to detect
Donovan> it if it happens.)

Exactly.  (In fact, the "reference implementation" has this property.)

-- 
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla