Re: [Scheme-reports] Formal Comment: R7RS 'eqv?' cannot be used for reliable memoization Mark H Weaver 12 Nov 2012 16:26 UTC

Alex Shinn <alexshinn@gmail.com> writes:

> Thank you for the well reasoned formal comment.  Note
>
> the formal comment period ended June 30,

For the record, lest anyone believe that I waited too long
to bring this up, I first raised this issue here on May 7:

  http://lists.scheme-reports.org/pipermail/scheme-reports/2012-May/002153.html

> and we've
> published a draft which many people are now reviewing and
> will be used for the ratification vote.  I will discuss with the WG,
> but it is unlikely we would consider such a change at this late time.

What is the purpose of allowing people to review a draft if you are
unwilling to fix serious flaws such as this?

> In particular, the issue you bring up was already voted on
> twice.  The definition of eqv? has historically been strongly
> contended, and there is simply no way to make everyone
> happy on this point.

I realize that there is contention over issues regarding NaNs (which I
left unspecified as in the current draft), and perhaps over the exact
wording, but who actively wants the current semantics which is not even
consistent with itself:

  (eqv? +0.0 -0.0) => #false  (if those are IEEE 754-2008 numbers)
  (eqv? +0.0 -0.0) => #true   (if those are MPFR numbers, or others)
  (eqv?    0  0.0) => #false

Who actively wants the above behavior?  It's a serious question.

Even you have admitted several times on irc that you don't like the
current definition.  Now we have new language that fixes these problems
and could be voted on.

Do you have a deadline that prevents you from issuing another draft?
Or have you simply decided that you don't _want_ to produce another one?

     Mark

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports