From what I've seen, Blob tags are not supported in Processbook and Datalink. Is it possible to use PI-AF and configure an AF attribute pointing to a specific bit in the Blob tag? If so, how is it done?
i think there is no way to configure something like a bit mask at the moment when you define your DataReference to an PI Tag.
It would also depend on the structure of your Blob (e.g. is it just an array of bytes, or is it mixed like you have an real-value and maybe an status-word and an error-bit). i did also check the analytics function reference (if available in your version of AF), but i could not find a function to mask specific bits).
at the EMEA User Conference i did hear about programmable Analytics which should give you similar functionality like ACE, but i think it's not yet available.
with this you might mask your bit with bit operators (bit operator AND...).
So a solution for for the moment would be the use of ACE
Anyway i hope for your requirements some other user knows a useful solution.
I was able to isolate the bytes from a Blob PI-Point in a AF Attribute with the following procedure:
1. Create a query in SQL Commander and SELECT the specific Blob tag / attribute
You should get something like this: TagName [86 26 46 56 81 11 23 23 56]
2. Use substr() function to isolate the desired character
Ex: SELECT ea.Name, substr(s.ValueStr,2,2) - The result is now only TagName 86
3. Create a View and export the result to this view
4. Head back to PI AF and link the view using PI OLEDB Provider
5. Use the Table Lookup Data Reference for you attribute
Basically you go through OLEDB enterprise to isolate the byte then you push the result to PI AF.
Franz and Marc-Olivier
As I thought, there is no straightforward solution...
Its good to see you on PISquare!
I was looking myself for content that is related to blobs and I cam across your post , so I may be a bit late...
The main thing about Blobs and this was already brought by Marc Andre is that if you are using them to store data, you will be responsible of transforming the data back into a format that your clients can support.
There are no clients currently that support the blob point type out of the box.
To me there is no big advantage of using a blob, its always better to have the data stored in a way that it is ready to consume by standard clients ( Datalink, Coresight, etc ) . And mainly because you don't want to prevent the users to do new things with the data ( that was not planned at the first place).
Perhaps your use case was something really specific though, that would be interesting to know more about it!
Have a nice week,
It's good to hear from you. Long time no see...
Thanks for this late answer. However, we used individual PI tags for the more significant data and SQL tables for the very slow and rarely used data. It works fine.
Thanks for the information, it's interesting to see how you could make things work.
Talk to you next time!
Retrieving data ...