[macstl-dev] Re: Question about sized booleans

[Glen Low]

Ilya:

On 20/07/2005, at 2:01 AM, Ilya Lipovsky wrote:

>> So that means that I was using your "select(...)" correctly. Now I  
>> will be investigating other venues of resolution. My compiler  
>> gives me this:
>>
>> error: cannot convert `stdext::valarray<float>' to `fd_set*' for  
>> argument `2' to `int select(int, fd_set*, fd_set*, fd_set*,  
>> timeval*)'
>>
>> even though I don't include any system headers.
>>
>
>
> Due to shortage of time (can't waste more than 2 hours on the silly  
> issue) I propose a workaround: how about if we rename "select" into  
> "sel?" Seems work fine for me. Even though "select" sounds more  
> understandable, "sel" is shorter to type. Btw, we never invoke  
> "exponent" -- only "exp." So it doesn't go against any conventions.
>
> What's your take on the proposition?

There is a conflict with an implicitly #included system header, but  
it should still be OK since the system select function is a C  
function with 5 parameters, while macstl's select has only 3  
parameters, so C++ overloading should do the right thing. Check that  
you're using the right # of parameters and select is on your include  
path.

You can also explicitly qualify select if you have problems i.e.  
stdext::select (a, b, c). It could be your compiler options are  
turning off argument-dependent lookup.

Shortening select to sel is dangerous. In altivec namespace, there's  
a sel function that corresponds exactly to vec_sel, and has different  
semantics:

altivec::sel (falsepart, truepart, bool)

A programmer might easily get the two sel's confused, especially when  
"using namespace".

Cheers, Glen Low


---
pixelglow software | simply brilliant stuff
www.pixelglow.com
aim: pixglen

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.pixelglow.com/lists/archive/macstl-dev/attachments/20050720/525df907/attachment.html