Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Mon, 5 Oct 2009 16:45:33 -0600

> I think Jeff is saying that current implementations of channel access do
> not expect the update size to change, but that there is nothing in the
> protocol over the wire to preclude such a change, merely the risk of
> breaking existing clients.

yes - clients, and much less likely (I will have to look at the code) old
versions of libca.

> (I'm going to guess that the CA wire protocol has no provision
> for such extensible settings, or does it?).

The protocol has a version number in it and different behaviors can be
assigned depending on the client knowing the server's version (and
visa-versa), but of course code size and maintenance headaches grow.

> I think that Jeff is saying precisely this, dynamic channel size doesn't
> exist because it's never been implemented in any channel access server.

correct

> Well clearly not, there's no mechanism there!  However, the Waveform
> record can do two things.  Firstly, send updates on changes to NORD!
> It's a bit remarkable that this doesn't happen already;  and secondly, of
> course, use dynamic channel size as soon as the CA server layer provides
> it.

Yes, for variable dynamic length updates to work the ca server would need an
interface for efficiently querying the current number of elements from the
database. It learns this currently from dbNameToAddr.

Jeff
______________________________________________________
Jeffrey O. Hill           Email        [email protected]
LANL MS H820              Voice        505 665 1831
Los Alamos NM 87545 USA   FAX          505 665 5107

Message content: TSPA

> -----Original Message-----
> From: Michael Abbott [mailto:[email protected]]
> Sent: Monday, October 05, 2009 10:56 AM
> To: Andrew Johnson
> Cc: Jeff Hill; [email protected]; [email protected]
> Subject: Re: Channel access dynamic array subscription update element
> count
> 
> On Mon, 5 Oct 2009, Andrew Johnson wrote:
> > Jeff,
> >
> > While you're responding on this subject could you definitively answer
> > Michael's original questions, which were:
> >
> > > Firstly, is it possible for the element count of a channel to change
> > > while the channel is connected?  I'm thinking in particular of the
> > > case when using camonitor (well, ok, I mean ca_create_subscription())
> > > to monitor a waveform PV.  Each event_handler callback is informed of
> > > the dbr count -- but as far as I can tell, it's always the maximum
> > > waveform size, even if NORD is smaller.
> >
> > > Is the fact that the update size doesn't change, ca_element_count()
> > > always reports the NELM, and subscribing with zero count gets the full
> > > wavelength, is this restriction fundamental to channel access, a
> > > limitation of the channel access server, or just something that hasn't
> > > been implemented in the waveform record?  This question becomes more
> > > interesting as we use waveform records for large images!
> 
> Let me try and understand Jeff's remarks to possibly make it easier for
> him to interpret my question.
> 
> By the options I listed:
> 
> 
> 1. Fundamental to channel access
> 
> I think Jeff is saying that current implementations of channel access do
> not expect the update size to change, but that there is nothing in the
> protocol over the wire to preclude such a change, merely the risk of
> breaking existing clients.
> 
> I like Jeff's suggestion of interpreting a zero size request as "give me
> what you've got, whatever size it is, and always send me the current
> size."  No doubt there are client applications that might have lazily
> allocated a fixed rx buffer but still specified zero size, and they'll
> break, but I suspect that's not so painful.  Point the finger at them,
> laugh, and fix them or tell them to set the CA_FIXED_SIZE option on their
> new servers (I'm going to guess that the CA wire protocol has no provision
> for such extensible settings, or does it?).
> 
> However, are there implementations of libca which are lazy in this way?
> Hopefully not!
> 
> 
> 2. A limitation of channel access server
> 
> I think that Jeff is saying precisely this, dynamic channel size doesn't
> exist because it's never been implemented in any channel access server.
> 
> 
> 3. Not implemented in Waveform record
> 
> Well clearly not, there's no mechanism there!  However, the Waveform
> record can do two things.  Firstly, send updates on changes to NORD!
> It's a bit remarkable that this doesn't happen already;  and secondly, of
> course, use dynamic channel size as soon as the CA server layer provides
> it.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System