[R6RS] Explicit vs. implicit names for record procedures

Michael Sperber sperber
Mon Jan 31 14:21:17 EST 2005

One of the record system design issues that's come up is the question
whether the record-type-defining form should explicitly list all
identifiers defined by it, or whether the identifiers are constructed
automatically (MAKE-<type>, <type>-<field>, <type>?) and so on.

I've been trying to observe my development habits over the past few
months a little bit, and have realized that I grep *a lot* for the
identifiers defined by record-defining form, and have come to
appreciate that most code written for Scheme 48 allows me to
successfully find these identifiers that way.  (SRFI 9 and SRFI 57
both also require all identifiers to be explicitly listed.)
Conversely, I've often stumbled upon not being able to find
identifiers defined by DEFINE-STRUCT in PLT easily.

Consequently, I'd be sad if R6RS's default record system would be one
where the identifiers are implicitly generated, because that would
mean most new Scheme code would contain implicitly defined
identifiers, and I wouldn't be able to grep for them easily.

Cheers =8-} Mike
Friede, V?lkerverst?ndigung und ?berhaupt blabla

More information about the R6RS mailing list