RE: pyepics problem with PV.put(wait=True)

["Mooney, Tim M." <mooney@aps.anl.gov>]

The underlying epics.ca module has the code required to ask for a value and wait for a reply:
    epics.ca.get(chid, ftype=None, count=None, as_string=False, as_numpy=True, wait=True, timeout=None)
    (set wait==True to wait for a reply)
but (from a cursory look at the online docs) it looks like this capability is not exposed either by epics.caget() or via PV().

Tim Mooney (mooney@aps.anl.gov) (630)252-5417
Software Services Group (www.aps.anl.gov)
Advanced Photon Source, Argonne National Lab


________________________________________
From: tech-talk-bounces@aps.anl.gov [tech-talk-bounces@aps.anl.gov] on behalf of Mark Rivers [rivers@cars.uchicago.edu]
Sent: Thursday, August 14, 2014 9:06 PM
To: Jameson Graef Rollins; EPICS tech-talk
Subject: RE: pyepics problem with PV.put(wait=True)

Hi Jamie,

I suspect you are seeing a race condition.

The pv class by default sets the auto_monitor parameter to True.  This means that a CA monitor is put on the PV and PV.get just returns the last monitored value.

When you do pv.put(i, wait=True) that waits for the record to be processed, and all other records that process as a direct result of that record processing.  However, it does NOT wait for all CA monitor callbacks to complete.

You are doing pv.get immediately after the pv.put, and I suspect that the CA monitor callback has simply not had time to run.  I am not that familiar with the way that multi-threading works in Python, so I am not sure if you need to do anything to allow the CA callback a chance to happen between your pv.put and pv.get.

I think you can eliminate the problem by setting auto_monitor=False when you create the pv object.

Mark


________________________________________
From: tech-talk-bounces@aps.anl.gov [tech-talk-bounces@aps.anl.gov] on behalf of Jameson Graef Rollins [jrollins@ligo.caltech.edu]
Sent: Thursday, August 14, 2014 7:18 PM
To: EPICS tech-talk
Subject: pyepics problem with PV.put(wait=True)

Hi, folks.  I'm seeing what might be a bug in the behavior of PVs in the
pyepics interface, or at the very least something inconsistent with
advertised behavior.

I'm trying to use the wait=True option to PV.put() to wait for process
completion, after which I want to log the put action.  Here's a basic
example:

#!/usr/bin/env python
>From epics import PV
pv = PV('T1:TEST-SYS_B')
for i in range(10):
    pv.put(i, wait=True)
    print pv.get()

However, what I'm consistently getting as output from the above is
something like the following:

0.0
0.0
2.0
3.0
3.0
4.0
6.0
6.0
8.0
9.0

Note that the output is not at all what is expected: on three out of ten
writes a pv.get() immediately following a pv.put(wait=True) fails to
return the newly updated value.  I see the same failures if I instead
set use_complete=True and check for put_complete==True before issuing
pv.get().

If the put() has actually completed, why is the get() still seeing the
old value?

Thanks in advance for any help or suggestions.

jamie.