• Robby (17/17) Jan 24 2007 is there a way I can do this in the module scope? I would have thought
• Jarrett Billingsley (42/57) Jan 24 2007 Nope, whatever2 is a function and as such can't be evaluated at
• Robby (8/82) Jan 24 2007 porting a couple of c libs over and got caught on that in a header,
• Frits van Bommel (5/8) Jan 24 2007 I'm pretty sure you can only assign it in the appropriate constructor:
• Jarrett Billingsley (3/11) Jan 24 2007 Yeah, that's right.
• Frits van Bommel (4/4) Jan 24 2007 Robby wrote:
• Robby (3/8) Jan 24 2007 if only computers were gps enabled and synced clocks due to location

Robby <robby.lansaw gmail.com> writes:

is there a way I can do this in the module scope? I would have thought 
that since everything is known at compile time it would have flew, do I 
need to go about it in a different manner, or?

enum whatever
{
	one,
	two,
	three	
}
int whatever2(int x)
{
	return 1<< x;	
}

const int a = whatever2(whatever.one);
int b = whatever2(whatever.one);

testcode.d(11): Error: non-constant expression (whatever2)(0)
testcode.d(12): Error: non-constant expression (whatever2)(0)

"Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:

"Robby" <robby.lansaw gmail.com> wrote in message 
news:ep7nbd$1r1i$1 digitaldaemon.com...
 is there a way I can do this in the module scope? I would have thought 
 that since everything is known at compile time it would have flew, do I 
 need to go about it in a different manner, or?

 enum whatever
 {
 one,
 two,
 three }
 int whatever2(int x)
 {
 return 1<< x; }

 const int a = whatever2(whatever.one);
 int b = whatever2(whatever.one);

 testcode.d(11): Error: non-constant expression (whatever2)(0)
 testcode.d(12): Error: non-constant expression (whatever2)(0)

Nope, whatever2 is a function and as such can't be evaluated at 
compile-time.  However, you can use a template instead:

enum whatever
{
    one,
    two,
    three
}

template whatever2(int x)
{
    const int whatever2 = 1 << x;
}

const int a = whatever2!(whatever.one);
int b = whatever2!(whatever.one);

Note that whatever2 can then not be used as a function; it is always 
evaluated at compile-time, so you can't do something like

void main()
{
    int x;
    din.readf("%d", x);
    writefln(whatever!(x));
}

because then the const int inside whatever2 will have a non-constant 
initializer!

If you want whatever2 to be a function, you'll have to initialize your 
globals differently:

int whatever2(int x)
{
    return 1 << x;
}

const int a;
int b;

static this()
{
    a = whatever2(whatever.one);
    b = whatever2(whatever.one);
}

It looks weird, declaring a as a const and then assigning it in the static 
this(), but that's perfectly legal -- you can assign to constant values 
once, but that assignment can happen anywhere. 

Robby <robby.lansaw gmail.com> writes:

Jarrett Billingsley wrote:
 "Robby" <robby.lansaw gmail.com> wrote in message 
 news:ep7nbd$1r1i$1 digitaldaemon.com...
 is there a way I can do this in the module scope? I would have thought 
 that since everything is known at compile time it would have flew, do I 
 need to go about it in a different manner, or?

 enum whatever
 {
 one,
 two,
 three }
 int whatever2(int x)
 {
 return 1<< x; }

 const int a = whatever2(whatever.one);
 int b = whatever2(whatever.one);

 testcode.d(11): Error: non-constant expression (whatever2)(0)
 testcode.d(12): Error: non-constant expression (whatever2)(0)

 
 Nope, whatever2 is a function and as such can't be evaluated at 
 compile-time.  However, you can use a template instead:
 
 enum whatever
 {
     one,
     two,
     three
 }
 
 template whatever2(int x)
 {
     const int whatever2 = 1 << x;
 }
 
 const int a = whatever2!(whatever.one);
 int b = whatever2!(whatever.one);
 
 Note that whatever2 can then not be used as a function; it is always 
 evaluated at compile-time, so you can't do something like
 
 void main()
 {
     int x;
     din.readf("%d", x);
     writefln(whatever!(x));
 }
 
 because then the const int inside whatever2 will have a non-constant 
 initializer!
 
 If you want whatever2 to be a function, you'll have to initialize your 
 globals differently:
 
 int whatever2(int x)
 {
     return 1 << x;
 }
 
 const int a;
 int b;
 
 static this()
 {
     a = whatever2(whatever.one);
     b = whatever2(whatever.one);
 }
 
 It looks weird, declaring a as a const and then assigning it in the static 
 this(), but that's perfectly legal -- you can assign to constant values 
 once, but that assignment can happen anywhere. 
 
 

porting a couple of c libs over and got caught on that in a header, 
cheers for that. haven't seen that error before and was curious, thanks!

didn't know about the static this() approach either, though I must admit 

correctly)

Thanks again!
Robby

Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:

Jarrett Billingsley wrote:
 It looks weird, declaring a as a const and then assigning it in the static 
 this(), but that's perfectly legal -- you can assign to constant values 
 once, but that assignment can happen anywhere. 

I'm pretty sure you can only assign it in the appropriate constructor:
static member --> static this() as member of same class/struct/whatever
non-static member --> this() [other parameter lists also allowed]
module-scope --> static this() at module scope

"Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:

"Frits van Bommel" <fvbommel REMwOVExCAPSs.nl> wrote in message 
news:ep7qcc$1vrq$1 digitaldaemon.com...
 Jarrett Billingsley wrote:
 It looks weird, declaring a as a const and then assigning it in the 
 static this(), but that's perfectly legal -- you can assign to constant 
 values once, but that assignment can happen anywhere.

 I'm pretty sure you can only assign it in the appropriate constructor:
 static member --> static this() as member of same class/struct/whatever
 non-static member --> this() [other parameter lists also allowed]
 module-scope --> static this() at module scope

Yeah, that's right. 

Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:

Robby wrote:
[snip]

By the way, it looks like your computer's clock is 2 hours ahead. (or 
timezone setting is wrong)

Robby <robby.lansaw gmail.com> writes:

Frits van Bommel wrote:
 Robby wrote:
 [snip]
 
 By the way, it looks like your computer's clock is 2 hours ahead. (or 
 timezone setting is wrong)

if only computers were gps enabled and synced clocks due to location 
during traveling, sigh.. maybe in a few years. thanks