Signal Processing Toolbox Major Bug Fixes
This document describes major bug fixes in this release.
Click on a problem area listed below to read how it has been fixed.
FDATool Memory Leak Fixed
Fixed Segmentation Violation in buffer Function
Fixes to cremez Function
Fixes to remez Function
Segmentation Violation Fixed in upfirdn resample
Using Large Data Sets with upfirdn and resample
FDATool Memory Leak Fixed
A memory leak while using the FDATool graphical user interface has been
fixed.
Fixed Segmentation Violation in buffer Function
Previously, a segmentation fault occured for the buffer
function with the nodelay
option when the overlap equaled
the length of signal. This has been fixed.
Fixes to cremez Function
A number of fixes have been made to cremez
, which has been
renamed to cfirpm
:
- Correct results for odd order highpass and bandstop filters are now
produced.
- Using the hilbfilt flag now produces correct results.
- Using the invsinc option uses a default gain argument if 'defaults' is
specified.
- Odd and even cases now return a real-valued result and even-ordered
cases do not cause an error.
- The
fresp
function, when called by cremez
,
defaults to 'none' if no default SYM is specified.
- Using a 'real' flag now produces a real-valued impulse response.
Fixes to remez Function
A number of fixes have been made to remez
, which has been
renamed to firpm
:
- The function's robustness has been improved. The design now converges
for many edge cases that used to have convergence problems. When the design
does not converge, an error message, with tips on how to modify the
specifications to try to attain convergencence, is displayed.
firpm
now requires odd-order symmetric filters to have
zero gain at the Nyquist frequency only if the last frequency point is
equal to 1. If your filter does not have zero gain at the Nyquist
frequency, you can either make the order even or use an 'h'
flag to prevent the filter order from being incremented and design a
Type 4 filter.
Segmentation Violation Fixed in upfirdn resample
There was a bug in upfirdn
, which manifested itself in
resample
, that caused segmentation violations. This bug has
been fixed.
Using Large Data Sets with upfirdn and resample
upfirdn
and resample
no longer cause a
segmentation fault when processing large data sets.