Would like ability to use the value of an attribute array values when doing name substitution like %@attribute1[0]%

I would like to be able to do a Rollup Analysis of Booleans attributes, without casting or converting the boolean. As Rollup Functions I would suggest: Logical AND, Logical OR and maybe Count of True/False.
It can be use to create a Combined Status or a Master Alarm of several Units.

By allowing for attribute templates to be referenced across element templates allows for increased flexibility in build AF hierarchies without the need for duplicating attribute templates. .

In AF, right clicking an attribute reveals an options called "change display UOM", but it's not doing what is expected.

To change the displayed UOMs (if, for instance, we want to see "km/h" instead of "m/h" in PI Vision), we need to do the following :
1. In AF, change the "source unit" UOMs to the "source" uoms
2. In AF, change the "Default UOM" to what we want to see displayed.

This process could be made a lot simpler if we could simply use the "change display UOM" feature instead, by right clicking the attributes.

Currently, the values of string tags are stored in full for each event, which is a waste of space for string tags that have the same values appearing frequently. To save space, please use dictionary encoding (AKA dictionary coding or dictionary compression) to store the values of string tags. The space saved for string tags that have many repeated values should hopefully offset the extra space used for string tags that have mostly unique values.

When backfilling tags from AF its very easy to update many archives. I would like the backup script to only require free space for archives that don't exist in the target folder. We have periodic incremental backup failures do to a lack of free disk space, even though almost all of the archives being backuped up will replace files that already exist in the target folder

Because event frame start and end time are stored as a bigint datatype in PIFD.dbo.AFEventFrame, the precision only seems to be to the nearest millisecond. This is a problem for sub-millisecond event frames. Why aren't event frame start and end times stored as DateTime datatypes?

It would be interesting to add the Replace function (like the one in StringBuilder) directly in AF Analysis

When looking at element attributes or event frame attributes in PI System Explorer I would like to see [string1, string2, string 3] for the value field of an array attribute, instead of "String Array" as we see today.

When you quickly check out the current values of an EF or element, it much easier to recognize the values you're looking for if you see (a preview of) it directly in the Values field for the attribute.

"Hiding" it under the "String Array" key word makes it more difficult to spot!

it will be useful to schedule an analysis with a specific Time Zone in order to be aligned with the local time and, especially, with the local setting for DST.

If Excel is installed as a Windows Store Application instead of a Desktop Application, the PI AF Client installer doesn't allow PI Builder to be installed. The reason for that is that it can't determine the Excel version through the registry.

PI DataLink faces the same problem, but there you have a workaround by editing the Setup.ini file. See this KCS: "PI DataLink setup kit does not detect Office that has been installed via Windows Store (220739)". This does not appear to exist for PI Builder.

Either a workaround or a genuine solution would be appreciated.

You should be able to urge EFs using more complex search criterias' than the ones provide in the latest PI Server 2018 SP3 (StartTime, EndTIme, EF Template).
Most important criteria would be EF categories, but also some honorable mentions:
1. EF category
2. Analysis Name
3. Primary Element Name
4. Duration
5. Others

When querying PI Points within PI Builder we see the changedate and creationdate properties show up in the format "yyyy-MM-dd hh:mm:ss" which is not the default format that Excel wants for filtering and sorting columns by date/time. However, we do see the proper format for these two properties shown after clicking any cell in the same sheet as the PI Point query and hitting "Retrieve". This yields the format "MM/dd/yyyy hh:mm" which is easier to sort and filter through.

Some interfaces such as PI Modbus allow the polling of a 64-bit signed integer value or a 64-bit unsigned integer value. Yet the database does not allow you to store these values in their native formats. You must store them as a 64-bit floating point but some precision is lost. Please add the ability to store these values in their native format.

It would be nice to have the possibility of setting a global resend and non-repetition interval.

At the moment PI API 1.x without Windows Integrated Security is installed as standard. For customers who want to use PI API 2.x with Windows Integrated Security an additional installation is needed. please include both APIs in install-pack and ask during installation what Version should be installed.