SWP 6.0.27 Arithmetic bug

jm's picture

I attach a sci file in which using Compute > Evaluate numeric in SWP 6.0.27 gives different arithmetic answers to what look like exactly the same division problem on-screen, in PDF and in TeX output.

It is deeply troubling for those of us who rely on the computation engine in SWP that this can happen. How can I ensure it does not happen again?

AttachmentSize
SWP6027ArithmeticBug.sci75.03 KB
pviton's picture

From the source text I can

From the source text I can see how the problem arose, but I've no idea how it got generated. If you look at the denominator of the first fraction, you'll see that it's entered as "1." [stuff] "5349", while the second (correctly) is "1.5349". So the first fraction is being interpreted as
1.5198/(1/5349)=0.000284.

But this doesn't address the problem that they both appear identically on screen, with no clue as to what is being computed; this is obviously a very serious issue.

Incidentally, I selected the top fraction, then did Compute -> Interpret to see if we could get some information (though I agree that this should not be necessary). Nothing happened. Is this feature working?