I had hoped for round-half-to-even instead of away from zero.

• This is IEEE 754 default rounding mode
  
• It is most common in engineering
  
• It is the way I teach my students

What is the motivation of the design decision of rounding half away from zero?

The step-in feature without breakpoints is great.
This is a major step towards a very convenient context-inspector.

IMHO these features could be useful:
- Context display at cursor location (what you would see if there was a breakpoint there), perhaps some execute-to-cursor button
- Reliable call-outs for dynamic formatted display of the context items (currently, sometimes the callout doesn't show up)

Martin, I'm recieving many e-mails regarding "rounding is unpredictable".
Personally I do not see significant benefits from round-half-to-even (because SMath Studio never knows source of the data it works with), but this, of course, has no sence.

Wrote

Martin, I'm recieving many e-mails regarding "rounding is unpredictable".
Personally I do not see significant benefits from round-half-to-even (because SMath Studio never knows source of the data it works with), but this, of course, has no sence.

The benefit is that if summing up large numbers of rounded numbers then you don't get a systematic bias.
In round away from zero this is true only if the distribution of the numbers is symmetric to zero.

Maxima implements round to even.

As the type of rounding seems to be a matter of taste I would recommend to make this a settings item, at least for result formatting. For explicit rounding one could have separate functions or an optional argument.


roundtest.sm (7 КиБ) скачан 88 раз(а).

The Text Region Utilities plugin fails to start





Best Regards

Carlos

Wrote

Martin, I'm recieving many e-mails regarding "rounding is unpredictable".
Personally I do not see significant benefits from round-half-to-even (because SMath Studio never knows source of the data it works with), but this, of course, has no sence.

====================================
Well, there is "rounding & rounding" ... most interesting:
http://mathworld.wolfram.com/RoundoffError.html

Nothing is absolutely true in this world.
1. NO math computations are true.
2. NO measurements are true.
3. NOT all minutes have 60 seconds ... What !

Yes "What" ... once in a while there is a Savant Lab [in Paris]
that adjusts the universal time ± 1 second.
I can survive the conventional ½ way up/down.

Cheers ... Jean.

Wrote

The Text Region Utilities plugin fails to start





Best Regards

Carlos

Updated, thank you!