Manta Interactive Ray Tracer Development Mailing List

Text archives Help


Re: [MANTA] unnamed in PointVector


Chronological Thread 
  • From: James Bigler <bigler@cs.utah.edu>
  • Cc: manta@sci.utah.edu
  • Subject: Re: [MANTA] unnamed in PointVector
  • Date: Tue, 24 May 2005 15:35:00 -0600
That did the trick. And just to make sure, I reversed the order of the returns to purposefully break and it did. This will also help the compiler not be stupid about make stack room (in case it decides it wants to).

Checkin coming soon.

James

Christiaan Paul Gribble wrote:
Placing the "typedef" keyword in front of the declaration may make those
warnings go away, but still allow the compiler to bail if Dim != 3.

C

James Bigler wrote:

This is exactly why I brought it up.  I can't see the real warnings from
the swarms of these ones.

James

Abe Stephens wrote:


It's also responsible for approximately 300 warning messages (yellow,
not red flags) if you turn on all of the warnings (which cmake or
xcode decided to do by default)..

Abe

Christiaan Paul Gribble wrote:


Looks like a compile-time check on the dimensionality of the vector to
me.  Anyway, it reminds me of Andrew's trick for the scratchpad size
that floated on the cs6620 list last spring:

template<typename T>
const T* HitRecord::getScratchpad(void) const {
  // Compile time check
  typedef char unused[(sizeof(T)<=MAX_SCRATCHPAD_SIZE)?1:0];

  return reinterpret_cast<const T*>(scratch);
}

In this case, if Dim == 3, char unnamed[1] will be a valid (but ignored)
variable, and if Dim != 3, char unnamed[0] will throw up a red flag...

C

James Bigler wrote:



What is the purpose of unnamed in PointVector.h?  It doesn't seem to be
referenced.  Here's an example from line 22:

  VectorT(T x, T y, T z) {
    char unnamed[ Dim == 3 ? 1 : 0 ];
    data[0] = x; data[1] = y; data[2] = z;
  }

Thanks,
James






Archive powered by MHonArc 2.6.16.

Top of page