#386 - notron - Tue Jan 07, 2003 4:15 am
Okay, I know I am a Newbie and this may be stupid, but..... it appears to me that G++ (devkitadv) is expanding 16 bits to 32 bits when it evaluates unary expressions. For example, I tested the following:
unsigned short mask = 0xffff;
if ((~mask) == 0)
{ (code turns screen blue) }
else
{ (code turns screen red)}
Now, the bitwise complement of mask should be 0x0000, but the above code will turn the screen red every time. If I embed an explicit typedef override in the code like this:
if ( ( (u16)~mask) == 0)
Then it properly evaluates to TRUE !
Okay, so tell me if me what gives here?
_________________
MysticX is The Defender
#391 - anli - Tue Jan 07, 2003 9:47 am
As far as I see, you cannot avoid this promotion to take place.
The typecasting is necessary, I think.
best regards
/anli
#415 - Burre - Tue Jan 07, 2003 4:39 pm
| anli wrote: |
As far as I see, you cannot avoid this promotion to take place.
The typecasting is necessary, I think.
best regards
/anli |
Yeah, I have to do a lot of typecasting (that shouldn't be necessary) in HAM using C++ as well. Especially when using pointers and adresses.
When using C lines like this works perfectly:
ham_StartIntHandler(INT_TYPE_VBL, &start_vblFunc);
But when using C++ it requires typecasting to (void*):
ham_StartIntHandler(INT_TYPE_VBL, (void*)&start_vblFunc);
Or else it will report an error (something like: cannot convert void (*)() to void*). Is it supposed to be that way?
_________________
"The best optimizer is between your ears..."
#448 - tom - Tue Jan 07, 2003 7:58 pm
[quote="Burre"]Or else it will report an error (something like: cannot convert void (*)() to void*). Is it supposed to be that way?[/quote]
C++ has much stronger typechecking than C, and void(*)(), a pointer to a function with no parameters and returning void is definitely not the same
as void*, so you need a typecast here.
So yes, this is supposed to be that way. Wether you regard this as a good or a bad thing is up to you =)
#456 - Burre - Tue Jan 07, 2003 8:35 pm
| tom wrote: |
| Burre wrote: | | Or else it will report an error (something like: cannot convert void (*)() to void*). Is it supposed to be that way? |
C++ has much stronger typechecking than C, and void(*)(), a pointer to a function with no parameters and returning void is definitely not the same
as void*, so you need a typecast here.
So yes, this is supposed to be that way. Wether you regard this as a good or a bad thing is up to you =) |
Well I don't mind really. Thanks for pointing out the difference. I'm used to C and have only coded in C++ a couple of months, so I'm not entirely used to all the differences yet. But yes I've noticed some differences in the use of explicit and implicit typeconversions.
_________________
"The best optimizer is between your ears..."