Re: [Discuss-gnuradio] Bug with set_taps in pfb_clock_sync

On Tue, Nov 24, 2015 at 3:42 PM, Richard Bell <richard.bell4@gmail.com> wrote:

I think I've come across a bug with the pfb_clock_sync blocks set_taps implementation. If I change a parameter in python that causes a change to the filter tap variable being fed to the pfb_clock_sync block, which then requires pfb_clock_sync.set_taps() to be called, the program will error out with the following:

Traceback (most recent call last):
  File "radio_devel/scripts/pcodes_multi_sim.py", line 484, in <module>
    main()
  File "radio_devel/scripts/pcodes_multi_sim.py", line 205, in main
    tb.set_rolloff(rolloff)   # same problem with pfb_filter
  File "/home/rbell/Documents/pcodes/radio_devel/scripts/pcodes_bpsk_stream_FEC_loopback_costas_nogui.py", line 184, in set_rolloff
    self.digital_pfb_clock_sync_xxx_0.set_taps((firdes.root_raised_cosine(self.nfilts_pfb, self.nfilts_pfb*self.samp_symb, 1, self.rolloff, self.nfilts_pfb*11*self.samp_symb+1)))
  File "/usr/local/lib/python2.7/dist-packages/gnuradio/digital/digital_swig.py", line 25452, in set_taps
    return _digital_swig.pfb_clock_sync_ccf_sptr_set_taps(self, *args, **kwargs)
TypeError: Required argument 'ourtaps' (pos 3) not found

Two examples of variables that cause this change are samp_per_symb or shaping filter rolloff factor. I only set these variables while the simulation is not running, i.e. before I call tb.start(). I don't know why setting the rolloff should break the pfb_clock_sync.

Rich

Setting the rolloff forces and update to the filter taps, so we have to reconstruct the filter and reload it into the filterbank. We made a change in how the loading process happens, so this looks like a pretty simple bug to fix how the update calls are applied. 

You should open an issue about this (or post a pull request with the bug fix).

Tom