Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Tue, 25 Feb 2014 12:26:51 +0100

On Monday, February 24, 2014 10:18:24 you wrote:
> On 02/24/2014 05:01 AM, Benjamin Franksen wrote:
> > There is no need to do large amounts of copying. You simply have to
> >
> >  allocate a new array before filling it with new data, instead of
> >
> > overwriting a previously allocated one. Then all the arrays can be
> >
> >  easily and safely cached/buffered/queued by reference. Indeed
> >
> > there is no need to *ever* copy such an array except, maybe, when
> > finally sending it over the network.
>
> Are you sure you're talking about features that appear in released
> versions of EPICS base? TTBOMK the support for double-buffering of
> arrays only appears in the unreleased 3.15 branch of Base, although
> I'm going to guess that Michael Davidsaver's Debian packages may have
> a back-port of that that functionality to 3.14.

Hi Andrew

I was arguing from principle, largely ignoring what's been implemented
and what not. Sorry, should have made that clearer. The point was (or
is) that I can see no reason in principle why costly copying would be
necessary to implement queued or buffered CA for array PVs. I have no
idea how deep the changes would be to actually implement this.

Cheers
Ben
--
"Make it so they have to reboot after every typo." ― Scott Adams

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System