Howdy:

The board brush strokes are easy.  I do not have enough patience, 
smarts, or time to do what Bob Cowdery has done in his fantastic erlang 
implementation.  On the other hand it does not fit with my vision of 
this and so far,  Frank agrees with what I have been saying so I am 
assuming it is substantially  his vision as well.

The idea is to have the erlang be the simplest possible thing,  as few 
lines as possible to get the job done reliably, robustly,  with remoting 
built in from the word go.

Cnodes are big in your future.  dttsp's sdr_core (as an example) will 
start up as a stand alone program and it will maintain its command 
structure.  Now  you use the simple cmdr script to send it commands. 
e_cmdr will be a C program that starts up and attaches itself to the 
TINY erlang server.  After it establishes its presence,  it will 
negotiate what it can provide and what it wants to receive.  So long as 
it is around, the erlang based hub will send all commands of the 
registered type to it and respond to all data requests of the type 
registered from e_cmdr by making a request  for this data from e_cmdr, etc.

The state held in the erlang radiocore will be the absolute minimum 
needed.  If the cnode disappears,  the hub will time out on it and drop 
it and  all of its sources and sinks from the list.

You might think of the, erlang cnode as working the exact same way Frank 
did the code in update.c BUT the "CTE" array will be dynamic and 
supplied upon initial connection, etc. in the radiocore.  It really will 
be TINY block of code and the smarts will be in the leaves.

I have had Cnodes running robustly on all my machines of every type for 
months.  I keep thinking that it is but a day's work to give everyone 
the e_cmdr as a template for how to make their connection to the hub and 
even to provide the simplest possible kind of GUI, something like 
Roger's little push button GUI  but interfaced through its own Cnode 
interface to the hub as a template for how to proceed.


Between having FOUR close relatives with life threatening conditions and 
being constantly on the road for work,  I just have not gotten it done.  
I apologize to everyone.  No one is more unhappy about this than me.  I 
need an uninterrupted week at home.  This week I am in Boston fighting 
the snow going to Cell processor class at Mercury and next week I am at 
VPI working on OFDM with GnuRadio folks for MY WORK.  I will catch a 
breath soon.

Bob


kd5nwa wrote:
> I downloaded the code from the SVN a couple of days ago and I'm 
> wondering what will the Radio core will do, I don't need a detailed 
> explanation just some broad strokes.
>
> At 05:15 PM 2/15/2007, you wrote:
>   
>> Walt:
>>
>> Frank, Eric, and I have specfically answered all of your questions about
>> the Linux support.  It is completely controllable under Linux now
>> especially for those who have no need for fancy radio GUI's.  For those
>> that need GUI's there ARE GUI's and more will be coming.
>>
>> Roger Rehr has painted the road with crayola crayon especially for you:
>>
>> http://www.nitehawk.com/w3sz
>>
>>
>> but the Linux code does not have the polished presentation of the 
>> Windows version of code AND IT DOES NOT SUFFER FROM ITS 
>> LIMITATIONS.  It is poised to take off now.
>>
>> There are several approaches to take such as dttsp-shell (available 
>>     
> >from Edson Pereira) or the java GUI done by John Melton or usSDR gui 
>   
>> done by Jonathan Naylor.  Frank and I are not interested in doing 
>> this GUI work.   We are interested in support anyone who wants to do 
>> the GUI work.
>>
>> Bob
>> N4HY
>> (coauthor DttSP with AB2KT)
>>
>>
>>
>>     


-- 
AMSAT Director and VP Engineering. Member: ARRL, AMSAT-DL,
TAPR, Packrats, NJQRP, QRP ARCI, QCWA, FRC. ARRL SDR WG Chair
"Taking fun as simply fun and earnestness in earnest shows
how thoroughly thou none of the two discernest." - Piet Hine

Reply via email to