Python vs. Perl, which is better to learn?

[James J. Besemer]

Andrew Dalke wrote:

> > I pondered for a while adding the necessary masking operations but
> > decided it was not sane.

I meant my first idea was to add the necessary masking ops to INTEGERS to
prevent overflow and I suspect you'll agree that would have been silly, even
though it might have worked.  I certainly did not mean to cast aspersions on
Python.

Then I tried using longs but it wasn't working right off, so I gave up.  It
was cheaper to write off the investment to date and switch to C, which free
off the web site.

> Here's the (completely untested and probably incorrect Python but
> close enough for the gist)

[Now with my bug-fixes]

>
> def encipher(v, k):
>   low32bits = 2L**32-1
>   y = v & low32bits
>   z = (v>>32) & low32bits
>   sum = 0L
>   delta = 0x9E3779B9L
>   a = k & low32bits
>   b = (k>>32) & low32bits
>   c = (k>>64) & low32bits
>   d = (k>>96) & low32bits
>
>   for i in range(32):
>       sum += delta
>       sum &= low32bits
>       y += (z << 4)+a ^ z+sum ^ (z >> 5)+b
>       y &= low32bits
>       z += (y << 4)+c ^ y+sum ^ (y >> 5)+d
>       z &= low32bits
>
>   return ( z << 32 ) | y
>
> Seems reasonable enough to me, and not very insane.

You're right.  Aside from a few bugs, I don't see why it wouldn't work.  Your
variables z, b and d needed to be shifted so the data is in the lower 32
bits.  Delta also needed to be long, as it affects the result of "sum+=".
Return result was undefined.  Since + is done before ^ your translation of
"+=" materially affected the computation.

I tried it with those changes and it appears to work.

Regards

--jb

--
James J. Besemer  503-280-0838 voice
http://cascade-sys.com  503-280-0375 fax
mailto:jb at cascade-sys.com