Menu
▾
▴
epics-pvdata-devel
|
From: Marty K. <mrk...@co...> - 2011-09-27 11:06:58
Attachments:
pvDataEtc.ppt
|
I have attached the first draft of a talk at the upcoming EPICS meeting at PSI. Comments? Marty |
|
From: Malitsky, N. D <mal...@bn...> - 2011-09-28 13:55:49
|
Hi Marty, It looks the present version of pvData inherited the previous problem. According to slide 7, timeStamp is used for two tasks: type definition and a name of the member in the user structure. Can we use two timeStamp's members in the same structure? -Nikolay -----Original Message----- From: Marty Kraimer [mailto:mrk...@co...] Sent: Tuesday, September 27, 2011 7:07 AM To: epi...@li... Subject: Talk for epics meeting I have attached the first draft of a talk at the upcoming EPICS meeting at PSI. Comments? Marty |
|
From: Marty K. <mrk...@co...> - 2011-09-30 14:08:03
|
On 09/28/2011 09:55 AM, Malitsky, Nikolay D wrote:
> Hi Marty,
>
> It looks the present version of pvData inherited the previous problem.
> According to slide 7, timeStamp is used for two tasks: type definition
> and a name of the member in the user structure.
>
> Can we use two timeStamp's members in the same structure?
>
At different levels in the same structure
structure top
timeStamp timeStamp // timeStamp for entire structure
...
structure input
int value
timeStamp timeStamp
In the pvIOC there is code that handles the top level timeStamp just
like in a V3 IOC.
There is no general purpose code that looks for or accesses lower level
timeStamps.
But there is nothing to prevent some future special code from using it.
Marty
> -Nikolay
>
> -----Original Message-----
> From: Marty Kraimer [mailto:mrk...@co...]
> Sent: Tuesday, September 27, 2011 7:07 AM
> To: epi...@li...
> Subject: Talk for epics meeting
>
> I have attached the first draft of a talk at the upcoming EPICS meeting
> at PSI.
>
> Comments?
>
> Marty
>
|
|
From: Malitsky, N. D <mal...@bn...> - 2011-09-30 14:18:08
|
>At different levels in the same structure
>structure top
> timeStamp timeStamp // timeStamp for entire structure
> ...
> structure input
> int value
> timeStamp timestamp
Following slide 7, the above structure should be written as
structure top
structure timeStamp // timeStamp for entire structure
long secsPastEpoch
int nanoSeconds
structure input
int value
structure timeStamp
long secsPastEpoch
int nanoseconds
Right?
-Nikolay
|
|
From: Marty K. <mrk...@co...> - 2011-09-30 14:34:37
|
On 09/30/2011 10:17 AM, Malitsky, Nikolay D wrote:
>> At different levels in the same structure
>> structure top
>> timeStamp timeStamp // timeStamp for entire structure
>> ...
>> structure input
>> int value
>> timeStamp timestamp
> Following slide 7, the above structure should be written as
>
> structure top
> structure timeStamp // timeStamp for entire structure
> long secsPastEpoch
> int nanoSeconds
> structure input
> int value
> structure timeStamp
> long secsPastEpoch
> int nanoseconds
>
> Right?
>
> -Nikolay
>
>
The metadata syntax allows both.
If
structure timeStamp
long secsPastEpoch
int nanoSeconds
has been defined than for metadata it is OK to say
structure top
timeStamp timeStamp
This can help make documentation more concise.
Marty
|
|
From: Malitsky, N. D <mal...@bn...> - 2011-09-30 14:36:44
|
>The metadata syntax allows both.
Great. Thanks.
-Nikolay
-----Original Message-----
From: Marty Kraimer [mailto:mrk...@co...]
Sent: Friday, September 30, 2011 10:34 AM
To: epi...@li...
Subject: Re: Talk for epics meeting
On 09/30/2011 10:17 AM, Malitsky, Nikolay D wrote:
>> At different levels in the same structure
>> structure top
>> timeStamp timeStamp // timeStamp for entire structure
>> ...
>> structure input
>> int value
>> timeStamp timestamp
> Following slide 7, the above structure should be written as
>
> structure top
> structure timeStamp // timeStamp for entire structure
> long secsPastEpoch
> int nanoSeconds
> structure input
> int value
> structure timeStamp
> long secsPastEpoch
> int nanoseconds
>
> Right?
>
> -Nikolay
>
>
The metadata syntax allows both.
If
structure timeStamp
long secsPastEpoch
int nanoSeconds
has been defined than for metadata it is OK to say
structure top
timeStamp timeStamp
This can help make documentation more concise.
Marty
------------------------------------------------------------------------
------
All of the data generated in your IT infrastructure is seriously
valuable.
Why? It contains a definitive record of application performance,
security
threats, fraudulent activity, and more. Splunk takes this data and makes
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2dcopy2
|
|
From: Korhonen T. <Tim...@ps...> - 2011-09-30 15:39:55
|
Marty Kraimer wrote: > > I actually have 40 minutes total. I plan to just let the end of power > point presentation lead into the demo. > The basic part of the demo should take at most 10 minutes. > If the power point presentation takes more than 20 minutes I can just > stop the demo after 10 minutes. > If on time more demo. > So Think maybe OK. > OK, then this is probably fine. >> *First slide: consider removing the duplicate statement about "no >> releases yet". >> > > OK it is gone. > >> *Introspection is a concept that is probably not that clear to >> everybody, although it would be too embarrassing >> to admit... would it be possible to: >> -explain what is meant by introspection >> -why it is a good idea (I think it is, but it would be nice to tell it >> to people. For instance that one does not need to transfer the >> metadata with the packets. And so on.) >> -how does it work in practice >> >> > > Let me prepare a slide and see if it helps. > Thanks for considering this. > > > Bob's talk is first. Also talks on services day before. > So I do not know how much more I can say to introduce the idea of services. > Perhaps refer to previous talks? > Suggestions welcome. > > I would perhaps just briefly refer to the other talks about services. I do not think you are short of material to go through ;-) >> *Property fields. A clever programmer made a remark to me recently >> that these _need_ to be standardized. It might be a good idea >> to clarify how these are to be handled - it is one of the big >> differences between CA and PVAccess/PVData. >> >> > > After my talks Bob has talk on "EPICS V4 Data Types". > His talk discusses Enum, Alarm, TimeStamp, Display. (Also others.) > pvData property explains pvData structures that provide the data for > these V4 Data types. > > Let me try changing slides to say this. > OK. > >> * I would consider leaving out the slides about introspection metadata >> syntax and the data metadata syntax. Sure, these are important >> but a programmer will look them up in the documentation. For somebody >> hearing first time about this, they do not contribute >> much to understanding. Or compress them somehow. >> > > Now I do not know what to do. > > I want to give people some idea of what pvData is other than just to say > that it supports structured data. > I am trying to briefly explain how is does it. > > What I could imagine to help people understand the idea would be to compare it to the present V3 record structures which are familiar to everybody. How does one create and populate a pvData structure and so on. I also think most people can appreciate something like an image structure where one adds the metadata about number of lines & rows, pixel depth, exposure, etc. These are things that one has to tweak into V3 records "somehow" without a good method to define them. This is also what many people immediately recognize as a need and an advantage. When people understand this, they start to wonder about how it is done. This is roughly the state where I am now. If anybody has other ideas it would be nice to hear about them. Timo >> * same as above for the slides about introspection and data >> interfaces. This (as the above) could be additional material, appended >> after the talk. >> >> > > Same comment. > If I do not give some idea of how pvData implements structured data I > think many people will be dissatisfied. > > Let me make some changes and then ask for more comments. > > Marty > > >> best regards, >> >> Timo >> >> >> Marty Kraimer wrote: >> >>> I have attached the latest draft of my talk next Friday. >>> Several good suggestions came from Bob. >>> One was to make bold new features that V4 has over V3. >>> >>> >>> Any more comments? >>> >>> Marty >>> >>> On 09/27/2011 07:06 AM, Marty Kraimer wrote: >>> >>>> I have attached the first draft of a talk at the upcoming EPICS >>>> meeting at PSI. >>>> >>>> Comments? >>>> >>>> Marty >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> >>>> All the data continuously generated in your IT infrastructure >>>> contains a >>>> definitive record of customers, application performance, security >>>> threats, fraudulent activity and more. Splunk takes this data and makes >>>> sense of it. Business sense. IT sense. Common sense. >>>> http://p.sf.net/sfu/splunk-d2dcopy1 >>>> >>> ------------------------------------------------------------------------ >>> >>> ------------------------------------------------------------------------------ >>> >>> All of the data generated in your IT infrastructure is seriously >>> valuable. >>> Why? It contains a definitive record of application performance, >>> security >>> threats, fraudulent activity, and more. Splunk takes this data and makes >>> sense of it. IT sense. And common sense. >>> http://p.sf.net/sfu/splunk-d2dcopy2 >>> >> > > > ------------------------------------------------------------------------------ > All of the data generated in your IT infrastructure is seriously valuable. > Why? It contains a definitive record of application performance, security > threats, fraudulent activity, and more. Splunk takes this data and makes > sense of it. IT sense. And common sense. > http://p.sf.net/sfu/splunk-d2dcopy2 > > -- Timo Korhonen PSI (Paul Scherrer Institut, http://www.psi.ch) CH-5232 Villigen PSI tel + 41- 56 3103262 fax + 41 - 56 310 5090 e-mail: tim...@ps... |
Thanks for helping keep SourceForge clean.
X