> Before such a version gets committed, I'd like to see it on Rietveld
> again.

Sure.  My original plan was to get the structural changes in first, and 
then worry about optimizations.

But now I think the x_divrem fix has to be considered a prerequisite for 
the 30-bit digits patch.  So I'll clean up the x_divrem code, include it 
in the basic patch and upload to Rietveld.  The other optimization (to 
multiplication) is more optional, potentially more controversial (it 
adds 60 or so lines of extra code), and should wait until after the 
commit and get a separate issue and review.

The x_divrem work actually simplifies the code (whilst not altering the 
underlying algorithm), as well as speeding it up.  Still, it's changing 
a subtle core algorithm, so we need to be *very* sure that the new 
code's correct before it goes in.  In particular, I want to add some 
tests to test_long that exercise the rare corner cases in the algorithm 
(which only become rarer when 30-bit digits are used).

I also want to understand Gregory's problems with configure before 
committing anything.

None of this is going to happen before the weekend, I'm afraid.