Hi
I were in the middle of using octave to generate input files for some VHDL testbench development I am doing. What I use it for is to make some calculations and save both input and output values as bit-vectors in a file.
This time, however, i had some negative numbers in both input and output values and I needed therefore to use the bitcmp function to calculate the two's-complement of the negative numbers.
Here I think that I maybe have found an error. The negative numbers are calculated simply as an bitwise XOR with the largest representable value in the unsigned datatype (i.e. 0xFF for 8 bit), whereas the correct two's-complement should be calculated as (a bitwise xor with the largest possible number) +1
I have changed this locally on my PC really easy, but I do not know where to file bugs and patches so that others can get information about this.