[R6RS] Case-sensitivity

William D Clinger will at ccs.neu.edu
Sun Feb 19 10:05:22 EST 2006


Matthew Flatt:
> But think about the message this sends to the community: the editors
> are so dysfunctional that they're *still* debating case-sensitivity.

Whereas the truth is that the editors are so dysfunctional
that we *haven't* been debating our tentative decision on
case-sensitivity?

> An definitive poll could have come at the beginning of the process, but
> it's too late. At this point, we should wait until we've produced
> results.

By then it's even later.  I'd rather get the case-sensitivity
debate out of the way beforehand, so the debate over a draft
R6RS can focus on more substantial matters.

Here's another compromise we should discuss:  Suppose the
R6RS does not specify whether Scheme is case-sensitive,
and leaves it up to individual implementations.  Then no
portable code can assume case-insensitivity, and no
portable code can assume case-sensitivity.  That is
pretty much the current de facto position, given that
several major implementations default to case-sensitive
in spite of the R5RS, and it is a major pain to get those
systems to work with a mixture of code that assumes case
matters and code that assumes case doesn't matter.

The effect of the compromise would be to require portable
code to be written with consistent case, while non-portable
code can be written according to any local convention with
respect to case.

We'd have to specify the case of identifiers like lambda
and quotient, but I imagine we'll say those are in lower
case.  We'd have to say that anyway if we stick with the
case-sensitive decision.

Will



More information about the R6RS mailing list