[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Discuss-gnuradio] Rearranging a running flow graph
From: |
Eric Blossom |
Subject: |
Re: [Discuss-gnuradio] Rearranging a running flow graph |
Date: |
Sat, 17 Mar 2007 13:09:25 -0700 |
User-agent: |
Mutt/1.5.9i |
On Sat, Mar 17, 2007 at 12:29:34PM +0100, Patrick Strasser wrote:
> Hello!
>
> I'd like to construct a demo for grc which can switch between two or
> more branches in a flow graph, like AM or FM: given a Signal, this would
> be modulated by an FM and an AM branch, then one is selected for a
> simulated channel (noise) and afterwards the signal is fed to either a
> FM or a AM demodulator. I want easily switch between the two behaviours,
> but am not shure about how to do this the Right Way.
I think the "right answer" will be available as soon as Johnathan
finishes the work on "disconnect" in the gr_hier_block2 stuff. This
will replace the existing hierarchical blocks, and will treat both
gr_blocks and hierarchical blocks as first class objects, able to be
connected and disconnected on the fly.
Johnathan, any estimate as to when this will be ready?
> I'd do this now with multiplicators, but that introduces for every
> branch a new block, and keeps the unused paths running, which is
> rousource hungry (I don't have that much CPU power).
Yes.
> An other way would be to do disconnects and connects in the flow graph.
> Is this a sensible way to go? Can I rearrange a running flow graph or do
> I have to stop before rearranging and start the flow graph afterwards?
> Josh: What do you think about integrating this into grc?
Yes, see above. The existing gr.hier_blocks don't handle disconnect well.
> The third possibility would be contruct a swtich block that takes n
> inputs and has a method to select the input by a number. Would this be
> possible an a way so that the not-to be consumed path temporarily
> discontinues processing? On continuation, the interrupted path would
> work with old data. Any way to flush this or are the buffers small
> enough to neglect this effect? I guess this would also be true for
> connect/disconnect.
>
> Patrick
Eric