Re: galculator should be included in Gnome

[Andrew Sobala <aes gnome org>]

On Sun, 2003-04-27 at 10:53, Simon Floery wrote:
> I'm the author of galculator and want to comment on the current discussion.
> 
> > There are HIG-compliance issues. 
> 
> That's right. At the moment galculator isn't very HIG conform. This is an issue
> I will address ASAP. 
> 

Please get rid of the "Hide menu bar" option. It's a "Please break me"
option. I couldn't get it back.

> There are also other features gcalctool has and galculator has not (financial
> mode, constants, user defined functions). At the moment I try to add features
> on request by users and the functional range of galculator will grow in the
> future for sure.
> 
> But talking about the future doesn't help at the moment, I agree.
> 
> > - uses libm; (...)  but doesn't do arbitary precision.
> 
> Following the discussion on the list I got the impression this is the
> strongest argument in favor of gcalctool. On the one hand I agree that accuracy
> is a characteristic to compare calculators objectively. 
> 
> But on the other hand the possibility to do calculations with upto 40 digits
> isn't a feature I expect from a desktop calculator. From my point of view a
> desktop calculator supports the user in doing (small) calculations on the paper
> (or even replaces that paper). If there are such big numbers I probably had to
> change to a bigger mathematical software package anyway. (In fact, I never got
> a feature request for supporting multiple-precision.)
> 

Some sort of standard form is very useful, though. Splatting the whole
number onto a display is just unwieldy.

How precise is libm? How precise is galculator internally? (Basically,
how many sig figs are accurate before weird rounding errors can start to
happen?)

> >From my point of view, galculator includes two very important features not
> mentioned in the current discussion:
> 
> Firstly, galculator supports RPN (reverse polish notation - a stack based
> approach on doing computations). Some users simply don't use a calculator not
> supporting RPN.
> 

Yeah, definitely a minority though. Although RPN can have its uses, it's
a specialist niche.

> Secondly, galculator pays respect to arithmetic precedence. Doing 1+2*3 you get
> 7 as the correct result in galculator (like gcalc), but gcalctool gives
> you 9 (1+2=3, 3*3=9). In my opinion one can't demand the user to think about
> arithmetic precedence and to resolve it with braces. Arithmetic precedence
> handling is a first step towards "Typing something in as it's written on paper
> should give the correct answer straight off".
> 

I'd say this is a very good reason for going with galculator, though it
conflicts with Anand Kumria's e-mail. It seems that galculator does know
about precedence, but still doesn't work to the "Typing something in as
it's written" principle. You have to do things like "1 + 10 log" to get
what should really be "1 + log 10". Easy-fix?

Anand Kumria's e-mail seems to be extremely inaccurate for both
calculators.

-- 
Andrew Sobala <aes gnome org>

"A freudian slip is when you say one thing but you mean your mother." -- unknown