[Discuss-gnuradio] Idea for gr-osmosdr

[Marcus D. Leech]

In developing simple_ra, I've run up against the issue that GRC doesn't 
really allow run-time reconfiguration of flow-graphs, but I have an emerging
  need to support slightly-different hardware configurations, involving 
either one or two input devices.

This is both for interferometer support, and differential radiometer 
support.  Now, I *could* laboriously maintain several different flow-graphs,
  and carefully maintain "operational consistency" between them, but 
I'd vastly prefer to support it all in a single flow-graph, with the ability
  to configure "features" at run time.

I think I can get most of the way there with existing functionality in 
GRC, but I'd like the ability to ask for a "fake" device (perhaps producing
  zeros, or gaussian noise).  That way, for single-device 
configurations, the 2nd channel can just be noise or zeros, which I just 
use a bit
  of flow-graph "algebra" to "do the right thing" with.

The gr-osmosdr source already provides a gaussian-noise stream when you 
misconfigure devices, but it's fixed at 1Msps.  It would be useful
  if this functionality could be made more "formal", and allow the 
effective sample rate to match the sample-rate of the "hardware participant"
  in a pair.

This isn't really the "right" way to go about this, but the "right" way 
requires a fair amount of roto-routering of GRC, or that I write my code
  in Python, and avoid GRC.  I'd rather not do that.


--
Marcus Leech
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org