As an AF administrator, I would like to understand the difference between an analysis that is stopped by an end user, or an analysis that has stopped but should likely be running. Currently when an analysis is stopped, we do not know if this is expected or not.

It would be to offer an additional 'status' like 'Stopped by user' or similar.

As someone who has migrated our calculations from PI ACE to Asset Analytics, we need to be able to annotate various calculation outputs in the same way we were able to with PI ACE. Please enable analyses to output annotations to various PI Tags in a similar fashion to how PI ACE allowed this.

Having more thorough and detailed messages for the Analysis service in the event logs, especially upon startup.

Environment details + Language settings that the Analysis service is using should be mentioned in the startup logs in order to better troubleshoot configuration issues without needing to go through a trial and error process.

When using an analysis variable as the output timestamp, it appears that the actual output timestamp is floored to the nearest second, shaving off any millisecond count. This is not a huge priority on our end, but I wanted to ask in case it was a quick and easy fix, as it would really help certain calculation streams to line up correctly with one another.

Our process is writing an array of strings into an AF attribute of the same type. This AF attribute may receive information of different size (index) in time.
We need to extract all the information contained in the string array (all the items) and store it in a PI tag of type string so it can be viewed in PI Vision or any other client.
Since there is no For-Loop function in ABA, we need a new array operation (for example GetStringContent()) to based on an event or periodically store this complete string in a PI tag.

The TagNum function seems to be the only function that requires a double quote as opposed to single quotes.
Many other TagXXX functions call for single quotes for its attribute or PI Point arguments.

When using the recorded values function in Analysis, Time weighted totals and averages assumes that the data is continuous and not discrete. This creates issues with data streams that are both discrete and require time weighting functions. Ideally a third calculation type needs to be added to Total, AVG and the like which is Time Weighted Discrete. This will adress this issue as is currently done in TagTot or TagAvg.

It would be nice if the PI Analysis Service could pick up a change in the Cache Refresh Interval of any AF Table being referenced by input attributes with a table lookup data reference. Currently, a service restart is required.

When you develop an expression it is very useful to use the preview function. Quite often you can discover and debug directly from this, but you are not abel to close down the window and check the code!
So either you memorise or you export the csv file ... which is quite cumbersom.
I suggest that you are allowed to minimse the preview window to revise your code so you can easily crosschek your findings. This small change would speed up development .

Please provide a 'GoTo' like Funciton for Analysis Expressions to skip expressions or maybe even go back to a previous Expression.
I realize you don't have line numbers per se, but if the Expressions operate in order, could you use the Variable Name?
This would be extremely useful for skipping a block of code under certain scenarios where it's just too difficult for a mile long nested if/then.
@ Kevin G., don't decline this, vote for it ;)

There are no number base conversion functions available in AF Analysis. There are many situations where I need to convert numbers between different 'number bases'. I frequently work with MODBUS devices and there are many times number base conversions are needed to properly get data.

Example: Convert decimal to hex, hex to binary, etc.

I test many of my functions using Excel before implementing them in Analysis. Excel has functions such as HEX2BIN, DEC2HEX, etc. I'm actually surprised that, after many years of AF/Analysis development, these type functions are not available. These are valuable functions that should be implemented.

We would like to be able to extract the data reference type (PI Point, None, Formula, Table Lookup, etc.) for an attribute using PI Analysis, similar to the TagType function for PI points.

See the attached picture. The picture also reveals other parts of PI System Explorer that are not translated. The picture and the idea for this suggestion come from https://pisquare.osisoft.com/s/feed/0D58W00006XNrBuSAL.

When input PI Tags are referenced in PI Vision 2020+ calculations, they do not automatically detect PI Tag name changes. For users with a lot of calculations, this can be extremely problematic in the event of a PI Tag name change. In the event of a name change, the following error is produced: [-12301] Performance Equation parsing error. Possibly updating the PE scheduler to reference the PointID instead would be beneficial.

During bulk recalculation or backfill, enhance the PI Asset Analytics engine to prioritize the calculation of analyses that act as inputs into higher level "daisy" chain analyses. This would address performance issues at large scale, as described in Known Issue "Autobackfilling or recalculating a large number of analyses with dependencies can be slow"

The ability to utilize three columns as input for interpolation from an AF Lookup Table to determine the result for a fourth column could be beneficial for predictions and analysis.

Currently, only linear and bilinear interpolation is possible from an AF Table Lookup.