
Boost : 
From: Fernando Cacciola \(Office\) (fcacciola_at_[hidden])
Date: 20021023 14:59:16
 Original Message 
From: "Guillaume Melquiond" <gmelquio_at_[hidden]>
To: "Boost mailing list" <boost_at_[hidden]>
Sent: Wednesday, October 23, 2002 3:45 PM
Subject: Re: [boost] two_pi or pi_twice?
> On Wed, 23 Oct 2002, Fernando Cacciola wrote:
>
> > > Or should it be the Interval library convention that puts the constant
> > > name first  an obvious plus point when making ordered lists  all the
> > > pi constants will come together.
> > >
> > > [details snipped]
> > >
> > FWIW, my own constants use *exactly* (word by word I think) the naming
> > convention you proposed.
>
> Would it be surprising if I said I prefer the way the constants in
> the Interval library are named? :)
>
> More seriously, I prefer using a sort of Polish inverse notation in order
> to have the main term in front of the name. For example, I think it's more
> homogeneous to have
>
This is what I expected... I figured that the Interval Library names came
from that notation.
> pi_half
> pi_twice
>
> rather than
>
> half_pi
> two_pi
>
I agree there is merit in the reverse notation... having the 'base' name
first is nice.
The problems I see are:
(a) the syntax for 'correct' postfix wordings is typically longer, and
perhaps uglier:
That is, I prefer 'two' over 'twice', 'four' over 'four_times', etc..
(b) correct postfix wording is not always available, so many times you have
to put a word that should go in front of the noun right after it, which
reads awkwards:
'pi_sqrt' (means sqrt(pi): sqrt_pi)
'pi_twice_minus' (means 2*pi: minus_two_pi)
> > > And while I am seeking views on names, do the following seem sensible:
> > >
> > > Explicit typed names
> > >
> > > pi_f // float
> > > pi_d // double
> > > pi_l // long double or pi_l ??
> > > // Potential ambiguity of long int and long double here!
> > > // C standard ISO 9899:1999 (a.k.a. C99) choses _l for long double
> > > pi_i // integer
> > > pi_li // long integer // or pi_l conflists with C99 ??
> > >
> > What about unsigned constants?
>
> Does that exist? The only unsigned constants that I can think of are the
> upper limits of the unsigned integral types. And I don't think the
> Constants library has to provide them.
>
I was not considering which type of constants actually existed. I was
thinking about possible combinations.
Even if there are no unsigned constants, I still prefer to separate basic
type (fp or int) from sizemodifier.
>
> > I'd rather use a 'type + (optional) modifier' postfix scheme:
> >
> > type:
> > f=floating point
> > i=integral
> >
> > modifiers:
> > s=short
> > l=long
> > u=unsigned
> >
> > pi_fs (float)
> > pi_f (double)
> > pi_fl (long double)
> > pi_i (int)
> > pi_il (long int)
> > pi_iu (unsigned int)
>
> If you already have 'pi_i', there is no need for 'pi_iu', is it?
>
Right, I was just typing combinatorily...didn't think about the actual
values.
>
> > > Interval limit static const constant variable names  add _l and _u
> > >
> > These will be easely confused with the 'type' postfix.
> > I'd use prefix instead:
> >
> > lower_pi_fl
> > higher_pi_fl
>
> I don't know if it's better to have long name. But I'm sure I wouldn't go
> for a prefix notation.
Because of the reverse notation scheme right?
> And I would also use suffixes with the same length:
>
> pi_fl_lower
> pi_fl_upper
>
> or
>
> pi_lower_fl
> pi_upper_fl
>
> Please note that 'lower' and 'upper' are the words used to refer to the
> bounds of an interval in the Interval library and I think it wouldn't be a
> bad idea to use the same notation in the two libraries.
>
Hmmm... My mind just neglected the word 'upper' to me :)
I really tried to find an opposite for lower with the same word length, but
upper didn't pop up in my head.
...it's too early to be this sleepy!
> Regards,
>
> Guillaume
>
Regards,
Fernando Cacciola
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk