Re: [Discuss-gnuradio] Problem with messages and msg_queue

[Trond Danielsen]

2007/5/10, Eric Blossom <address@hidden>:
> On Thu, May 10, 2007 at 05:02:21PM +0200, Trond Danielsen wrote:
> > 2007/5/10, Eric Blossom <address@hidden>:
> > >On Thu, May 10, 2007 at 10:24:04AM +0200, Trond Danielsen wrote:
> > >>
> > >> Ok, I read from pkt.py that to finish a packet, I just append
> > >> gr.message(1) at the tail of the message queue, but is there a simple
> > >> way to do this if I just want to send packet of fixed size, where the
> > >> payload comes from a flow graph?
> > >
> > >Hi Trond,
> > >
> > >The gr.message(1) isn't to end a packet, it's to tell the scheduler
> > >that there's no more data coming.
> > >
> > >I'm not exactly sure I'm following what you are trying to do.
> > >However, I think that if you follow through the logic in
> > >gnuradio-examples/python/digital/tunnel.py you'll see a way to send
> > >and receive packets (fixed or variable length) between python and the
> > >flow graph.
> > >
> > >Eric
> > >
> >
> > Okay, I will try to describe my problem a little better. I have a
> > block that estimates some properties on a received signal: Frequency
> > and phase. Lets call the block acquisition. The acquisition block
> > outputs the estimates at 1kHz. Now what I want to do is grab these
> > estimates and use them to update the center frequency and code delay
> > in another block, lets call it tracking :).
> >
> > I have tried to use probe_signal_f, but this does not work because the
> > acquisition and tracking block must be synchronized.

First of all, thank you for you patience with all the questions from a
simple user like me! I wouldn't have got very far without the support
from this mailing list.

> The easy way is to put them in the same block (Yes, I know that wasn't
> the answer you were looking for.)  Or you could try using a
> message queue and messages between them.  You don't have to use
> gr_message_sink.  You could have your estimator block create mesages
> (of whatever format it likes) and insert them into a message queue
> that was handed to it, say in the constructor.

Indeed it would easier if everything was in one block, but I should
mention that the acquisition block is defined at python level, and is
currently not written in C++. I would really like to keep it at this
level, both because of simplicity and for the flexibility it adds.

I do believe that the problem that I address is a fairly general one,
and something that might be easier with the new mblock. But from what
I could see, mblocks are currently only available from C++, or am I
missing something?

--
Trond Danielsen