[racket-dev] Sublinear functions of superfloat numbers

At Tue, 03 Jul 2012 08:37:24 -0600, Neil Toronto wrote:
> Anyway, I've decided to regard `log' (with huge rationals) and `sqrt' 
> (with perfect squares) as anomalies, because I'm finding more examples 
> that don't work. [...]
> 
> It's actually a bit of a relief that I don't have to worry about this.

That sounds right to me.

At Sun, 01 Jul 2012 18:00:31 -0600, Neil Toronto wrote:
> Apparently, argument reduction only happens to reals, and only in a few 
> functions. (But I'm glad it does, because the plot library uses `log' to 
> format huge numbers.)

I think the general rule, if you could call it a rule, has been "try
harder when it seems useful for something".