Why can't static const float class members be inititalized in the class?

[Martin Rennix]

Eg

class Foo
{
static const float f = 0.0f; // error
static const int i = 0; // ok
};

What's the reasoning behind this seemingly arbitrary limitation?

Martin

[Victor Bazarov]

Martin Rennix wrote:

class Foo
{
static const float f = 0.0f; // error
static const int i = 0; // ok
};

What's the reasoning behind this seemingly arbitrary limitation?

I would guess portability, but don't quote me on that.

Ask in comp.std.c++ -- they hold/discuss the reasoning behind any
decisions that made it to the Standard.

V
--
Please remove capital 'A's when replying by e-mail
I do not respond to top-posted replies, please don't ask

[Ron Natalie]

Martin Rennix wrote:

Eg

class Foo
{
static const float f = 0.0f; // error
static const int i = 0; // ok
};

What's the reasoning behind this seemingly arbitrary limitation?

Martin

The compiler is only required to know how to do integer math at compile
time.

[Martin Rennix]

On Apr 18, 8:41 am, Ron Natalie <[e-mail]:

MartinRennixwrote:

Eg

class Foo
{
static const float f = 0.0f; // error
static const int i = 0; // ok
};

What's the reasoning behind this seemingly arbitrary limitation?

Martin

The compiler is only required to know how to do integer math at compile
time.

Is that because in the "olden days" you couldn't guarantee that a FPU
would be available, or the equivalent software floating point library?

[James Kanze]

On Apr 18, 5:20 am, Martin Rennix <[e-mail]:

On Apr 18, 8:41 am, Ron Natalie <[e-mail]:

MartinRennixwrote:

Eg

class Foo
{
static const float f = 0.0f; // error
static const int i = 0; // ok
};

What's the reasoning behind this seemingly arbitrary limitation?

The compiler is only required to know how to do integer math at compile
time.

Also, the fact that there is no need for it. It's really a
hack, which was added for integers is because certain uses
require an integral constant expression.

Is that because in the "olden days" you couldn't guarantee that a FPU
would be available, or the equivalent software floating point library?

It's because even today, cross-compilers exist, and floating
point results depend on the actual floating point being used.
Something like 2. * 3.14 will have different values on different
machines. And the standard doesn't impose emulation of the
target floating point hardware.

--
James Kanze (GABI Software) email:ja*********@gmail.com
Conseils en informatique orientée objet/
Beratung in objektorientierter Datenverarbeitung
9 place Sémard, 78210 St.-Cyr-l'École, France, [phone]

[Ron Natalie]

Martin Rennix wrote:

>
Is that because in the "olden days" you couldn't guarantee that a FPU
would be available, or the equivalent software floating point library?

I think it was primarily there for crosscompilers. Integer math is
pretty easy to do universally, floating point is a bit more involved.
Not everybody uses the IEEE fp math.