Does the web service return the status information as a seperate field, or is the status information substituted for the actual data when it occurs?
Thanks Rhys, I'll double check.
Apparently we've tried both. I'm going to come clean - I'm not the developer. I'll try and encourage him to join this thread.
We've tried Nulls and text in the value column and +ve and -ve values in the status column.
Has anyone successfully got this to work?
It's in the documentation so it 'should' work.
Try always setting the status to 0 even when you want gaps in the data i.e. when you have "No Data" as the value being returned.
You can test your data in Excel with Datalink as a start - preview the data in RtWebParts/RtBaseline from you webservice, copy the data table into Excel and create a trend based on the data.
OK had a little play and a bug in the Trend object suddenly came back to me...do not use "No Data", apparently the bug (can someone from OSI confirm) is that No Data gets evaluated to the system digital state "Calc Failed" and it is "Calc Failed" that causes all data points not to be drawn - you will notice the trend scale is drawn as 0 -> 1.
So use "Bad" instead. I did this using a calculation DS, using "No Data" meant no values were drawn but switching to "Bad" was OK (see below):
if 'sinusoid' > 96 then 'sinusoid' else DigState("Bad")
Hope this helps....
Just so it is clear, above is screenshot of RtTrend.
That's pretty interesting Rhys and worth a punt, thanks.
The developer came back to us in the end. Apparently if you mimic an ACE webservice it works just dandy, as expected. You have to name the web service 'PIACEWebService'. You also need to behave like a PI ACE web service and return the right columns.
We've yet to wrap our services up and have them appear like this, but we'll keep you posted. It's an enormous faff I have to say, but at least we appear to have a workaround for the moment.
Are you using a class to return your values? which class?
We've not got any joy with tech support so far
We understand your issue and we'll work together to find a viable solution. That said, please allow me to clarify a few things here:
Technical Support is meant for regular use of our products, including reporting of bugs and enhancement requests such as the one you made (which resulted in PLI 19137OSI8) - essentially, any kind of "PI Administration" and "End User" type questions. By reading the call, it seems like they did a really great job in filing in that enhancement request and helping you as much as they could, before to refer you to vCampus.
And that's where we come into the picture That's precisely why vCampus was created: to encourage collaboration and share knowledge around integration as well as programming and designing solutions around PI products. That is, stuff that is not "out-of-the-box", that cannot be done natively with our products - which is what you're facing here... (PLI 19137OSI8 willl get solved but most likely later)
So, again, let's work together to solve this!
So when you expect a gap in the data are you supplying the appropriate status and "No Data" as the value? This has always been the trick for ProcessBook trends but not tried it for RtTrend although you should be able to replicate this using a PI Calculation Dataset (let me have a look).
There is a bug with the Trend object when a System Digital State is used, it will knock out the Trend Scales - maybe the bug extends to not displaying data.
When you preview the data I assume it all looks fine, it is just RtTrend not displaying it correctly..?