I've been working on simplified examples of grouping (and also explaining the first metering example that comes in the .ZIP). You hit me at a good time since I was struggling whether to build a Lua example or come up with a different grouping pattern. I'll work on your case and draw up a solution. GroupByID might work or it might be easier to do this with Lua script.
If I can restate your problem structure:
Your Origin database where all your tags are read in are broken down by a single Site element, each site has a long list of attributes, each with a PI Point reference to each tag that's on that site.
|- SITE1_LIY_7050 >> \\piserver\SITE1_LIY_7050
Then your target database that is transformed would look more like this
| - Total Volume >> \\piserver\SITE1_LIY_7050
| - Level Cell A >> \\piserver\SITE1_LI_7050A
| - Level Cell B >> \\piserver\SITE1_LI_7050B
| - Total Volume >> \\piserver\SITE1_LIY_7051
| - Level >> \\piserver\ITE1_LI_7051
Is that about right for your problem statement?
I apologize for falling off the face of the plaent. I haven't been monitoring this and forgot all about it. I believe I met you at PI World at the AFTransformer talk you gave (I was the guy who gave the glowing unsolicited AFTransformer endorsement and asked all of the Emacs questions at the end).
Your restatement of my problem is spot on. Currently I'm doing this by making a bunch of XML transformer files and running AFTransformer multiple times (about 21 passes) to build up the rich hierarchy at the end of the day. I tried to send a zip file of my work over to Dwaine, but the firewall/antivirus software stripped it out. The other approach I was contemplating was to read multiple shapes in and then try to write them out, but I got terribly confused and just went with the simple multi-pass approach even though it's a bit slower. With that said I can convert about 30-40 sites in about 15-20 minutes. This is about 2500 tags total (about 60-70 tags per site).
I'm currently working on other projects and have had to tabled this for now. I have our first cut of the new database done and playing with it using PIVision every now and then. I will definitely revisit this in the future and will continue to use AFTransformer.
Thanks and great hearing from you,
Good to hear Matthew!
We've definitely seen several instances of doing multi-pass runs of AFTransformer to piece together a composite AF Database, and also the reverse case (segmentation of a complicated DB into simplified department/component DBs).
If it works... it works! :-)
Thanks, Chris. It's nice to know we're not doing something that's just insane. Yeah, it works and we can get the job done. Things are easy to modify and having simple steps in the process makes troubleshooting much easier.