guruofquality
commented
11 years ago This is the cause of the failures in mpsk snr block. I dont think this negative indexing is ok, even when history is set.
int
digital_impl_mpsk_snr_est_svr::update(
int noutput_items,
const gr_complex *in)
{
for (int i = 0; i < noutput_items; i++){
double x = abs(in[i]);
double x1 = abs(in[i-1]);
double y1 = (x*x)*(x1*x1);
d_y1 = d_alpha*y1 + d_beta*d_y1;
double y2 = x*x*x*x;
d_y2 = d_alpha*y2 + d_beta*d_y2;
}
return noutput_items;
}
After the buffer re-factor and some gr_block wrapper tweaks, the problem qa tests seem to be passing. However there is still this fishy-ness surrounding these blocks. And it appears to be random. It could be a GRAS issue, a gr block wrapper issue, or just something funny about the blocks.
The blocks do have a few things in common: Relatively new, same author, perhaps use of history, relative rate...