If I use AF Analysis to output a calculation to an attribute , but do not write back or historize as Pi tag, it becomes too slow to trend this attribute in Pi Vision for more than a few months.
For certain KPIs and use cases we would like a quick view of long term trends [example 5 years]. In this case a daily sample of weekly sample would suffice. If I use Pi-datalink in excel and use the same attribute I can use "sampled data" function to quickly generate such a table in excel [takes me less than 20 seconds] for past 5 years with daily or weekly samples.
Why can't this neat trick of "sampled data" which probably uses some array function tricks for high speed be added to Pi Vision to make such long term trends. Currently the only workaround is to create a Pi tag for all such attributes and historize them to last 5 years.