1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 <2022> 2023 2024 2025 | Index | 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 <2022> 2023 2024 2025 |
<== Date ==> | <== Thread ==> |
---|
Subject: | Re: NTNDArray with attributes using QSRV group Pas |
From: | "Veseli, Sinisa via Tech-talk" <tech-talk at aps.anl.gov> |
To: | "tech-talk at aps.anl.gov" <tech-talk at aps.anl.gov>, Joao Paulo Martins <JoaoPaulo.Martins at ess.eu>, "Kasemir, Kay" <kasemirk at ornl.gov> |
Date: | Wed, 2 Nov 2022 15:40:50 +0000 |
Hi,
Although this does not help with your problem, perhaps it is worth looking into as possible alternative approach for serving images:
>>> import pvaccess as pva
>>> attr = pva.NtAttribute('x', pva.PvInt(3))
>>> attr['timeStamp'] = pva.PvTimeStamp(time.time())
>>> nda = pva.NtNdArray()
>>> nda['attribute'] = [attr]
>>> nda['timeStamp'] = pva.PvTimeStamp(time.time())
>>> s = pva.PvaServer('image', pva.NtNdArray())
>>> s.update('image', nda)
>>>
$ pvget image
image epics:nt/NTNDArray:1.0
union value
(none)
codec_t codec
string name
any parameters
(none)
long compressedSize 0
long uncompressedSize 0
dimension_t[] dimension
int uniqueId 0
time_t dataTimeStamp <undefined>
long secondsPastEpoch 0
int nanoseconds 0
int userTag 0
epics:nt/NTAttribute:1.0[] attribute
epics:nt/NTAttribute:1.0
string name x
any value
int 3
string[] tags []
string descriptor
structure alarm
int severity 0
int status 0
string message
structure timeStamp
long secondsPastEpoch 1667401566
int nanoseconds 915819406
int userTag -1
int sourceType 0
string source
string descriptor
alarm_t alarm
int severity 0
int status 0
string message
time_t timeStamp 2022-11-02 10:06:28.768 -1
long secondsPastEpoch 1667401588
int nanoseconds 768045902
int userTag -1
display_t display
double limitLow 0
double limitHigh 0
string description
string format
string units
Also note that pvapy has some useful utilities available for this:
>>> from pvapy.utility.adImageUtility import AdImageUtility
>>> image = AdImageUtility.generateNtNdArray2D(imageId=1, image=np.array([[1,2,3,4,5],[11,12,13,14,15]], dtype=np.uint8))
>>> print(image)
epics:nt/NTNDArray:1.0
union value
ubyte[] ubyteValue [1,2,3,4,5,11,12,13,14,15]
codec_t codec
string name
...Best Regards,
Sinisa
--
Siniša Veseli
Scientific Software Engineering & Data Management
Advanced Photon Source
Argonne National Laboratory
sveseli at anl.gov
(630)252-9182From: Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Kasemir, Kay via Tech-talk <tech-talk at aps.anl.gov>
Sent: Wednesday, November 2, 2022 9:28 AM To: tech-talk at aps.anl.gov <tech-talk at aps.anl.gov>; Joao Paulo Martins <JoaoPaulo.Martins at ess.eu> Subject: Re: NTNDArray with attributes using QSRV group Pas
I think this shows that the "info(Q:group, ..." approach may not be the one, only and final way to assemble information from records into custom PV Access data structures.
Info tags can be useful to configure a simple aggregation of records into a combined data type. But if we then want to control the exact naming and layout of the resulting structure, add details about the order of timing, time stamps, .. we're starting to invent
a new programming syntax. At that point, maybe it's better to fall back to an existing programming syntax like python or lua. We can already create custom PVA data structures from python, maybe tying that into the IOC's record processing is a better approach
for creating fully custom data structures?
From: Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Joao Paulo Martins via Tech-talk <tech-talk at aps.anl.gov>
Sent: Wednesday, November 2, 2022 10:04 AM To: tech-talk at aps.anl.gov <tech-talk at aps.anl.gov> Subject: [EXTERNAL] NTNDArray with attributes using QSRV group Pas Hello,
I’m trying to instantiate a NTNDArray with some NTAttributes in my IOC using QSRV group PV definitions, as the example in pva2pva (https://github.com/epics-base/pva2pva/blob/master/iocBoot/iocimagedemo/image.db ).
If I set the database as the following:
record(waveform, "$(P)ArrayData") { field(FTVL, "USHORT") field(NELM, "262144") info(Q:group, { "$(N):Array":{ +id:"epics:nt/NTNDArray:1.0", "value":{+type:"any", +channel:"VAL", +trigger:"*"}, "":{+type:"meta", +channel:"SEVR"} } }) }
record(mbbi, "$(P)FOO") { info(Q:group, { "$(N):Array":{ "attribute[0].value":{+type:"any", +channel:"VAL"}, "attribute[0].timeStamp":{+type:"meta", +channel:"TIME"}
} }) }
the resulting NTNDArray PV will have the following structure in its attribute[0] field:
structure structure timeStamp alarm_t alarm int severity 0 int status 0 string message NO_ALARM time_t timeStamp long secondsPastEpoch 1667394506 int nanoseconds 335496073 int userTag 0 any value double 0
The timeStamp values of the top structure are correctly taken from the “ArrayData” record. However, for the “attribute[0]” field, the timestamp values are taken from the “FOO” record, but the structure of the NTAttribute contains two other structures (alarm and timeStamp) inside the “top” timeStamp structure.
The question is: how one can set the “timeStamp” structure of the attribute[] taking the TIME information from a different record than the one used to get the array value?
Thanks in advance!
João Paulo Martins European Spallation Source ERIC
|