user-defined types

[alexshinn]

Do we support any means of creating disjoint user-defined types, such
as in SRFI-9, SRFI-99 or the R6RS record system?

[cowan]

I support providing the syntactic portion of SRFI 99 only. This is a very new SRFI, but it is completely backward compatible with SRFI 9, which is the most popular of all SRFIs (according to the documentation, only Chez among the major implementations does not support it). The extensions include single inheritance and (optional) implicit naming, along with succinct abbreviations for specifying whether a field is immutable or mutable.

I do not support, I reject, I am altogether against the standardization of R6RS records by WG1. (Or WG2, for that matter.) R6RS argues that compilers can make them more efficient than SRFI-9-style records, but SRFI 99 refutes this position.

[kumoyuki]

Yes. Disjoint programmer types are a must. Ideally, the whole type model should be seamless from the set membership predicates through to an effective type algebra. It would be nice to reify some form of type tags/descriptors across all types, as well as a limited pattern matching.

[cowan]

The WG voted to adopt SRFI-9 as part of the core.

[alexshinn]

reverting procedural/syntactic split

[cowan]

WG1 voted to accept SRFI 9.