Experimental Physics and
Industrial Control System

Juan F. Esteban Müller via Tech-talk <tech-talk at aps.anl.gov> · Wed, 1 Jul 2020 18:21:51 +0000

Thanks for the info. I don't fully understand the issue... Is there any workaround? Or in which cases the "dec" filter works?

Cheers,
Juan

On 2020-07-01, 18:22, "Michael Davidsaver" <mdavidsaver at gmail.com> wrote:

    On 7/1/20 9:09 AM, Juan F. Esteban Müller wrote:
    > Hi Michael,
    > 
    > Thanks for your reply.
    > 
    > I was suspicious, because I had the feeling I used it in the past... Can you maybe then give me a clue why I can't make it work in a simple example?
    > 
    > This is my db file:
    > record(calc, "Test:counter") {
    >     field(INPA, " Test:counter")
    >     field(CALC, "A+1")
    >     field(SCAN, "1 second")
    > }
    > 
    > And I run it using softIocPVA.
    > 
    > If I do
    > camonitor 'Test:counter.{"dec":{"n":2}}'
    > I get every second update, while if I do
    > pvmonitor 'Test:counter.{"dec":{"n":2}}'
    > I get nothing. Funny enough, if I do
    > pvmonitor 'Test:counter.{"dec":{"n":1}}'
    > then I get every update...
    > 
    > I am using base 7.0.4, but I also tested 7.0.3.1, which is the one I used in the past when testing the filters.

    Oh, right.  With the 'dec' filter you're seeing the effects of:

    https://github.com/epics-base/pva2pva/issues/32

    Specifically:

    > QSRV assumes that the first DBE_PROPERTY update will never be dropped.
    > If such an update is dropped, the subscription stalls.

    I don't have a full solution to this yet.

    Part of https://github.com/epics-base/epics-base/pull/71
    includes a change to Base which will allow filters to distinguish
    value and meta-data updates.  Of course this won't necessarily
    help with older Base versions.

    QSRV could also be changed to not wait for the initial meta-data
    update.  The result would be a double initial update for most new subscribers.
    There is nothing really wrong with this, I just found it annoying.

    > Cheers,
    > Juan
    > 
    > On 2020-07-01, 15:39, "Michael Davidsaver" <mdavidsaver at gmail.com> wrote:
    > 
    >     On 7/1/20 2:59 AM, Juan F. Esteban Müller via Tech-talk wrote:
    >     > Hi all,
    >     > 
    >     >  
    >     > 
    >     > I would like to know if there is anything equivalent to the CA decimation filter in PV access.
    >     > 
    >     > https://epics.anl.gov/base/R7-0/4-docs/filters.html#Decimation-Filter-dec
    >     
    >     These filters are implemented at the database level, and so work both with CA and PVA
    >     (via. the RSRV and QSRV server implementations).
    >     
    > 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System