When working with Spectre files, you can use certain categories of functions in all the file types: build, cplan, dive, measures, and qvset. These categories include Conversion Functions, Date and Time Functions, Logic Functions, Math Functions, and String Functions.
You cannot determine where other functions work by file type alone—instead, examine the context within the expression. For instance, you can use `value("X")` whenever the context involves only one record. If there are multiple records, you need to use `sum(value("X"))` or `info(value("X"))`, or another Summary function that processes rows of data to return one result.
Most Summary functions are meant to be used on multiple records to produce a single value. The expression argument they take runs on one row at a time. So the `value()` function, which only works on one row, can be used inside those Summary functions. Use this technique with the following Summary functions: any, average, first, geomean, harmean, info, kahan_sum, kurtosis, last, max, median, min, mul, skewness, stddev, and sum.
On the other hand, functions such as `top()` or `parent()` take an argument that is meant to be run on multiple records, so `value()` does not work. Other functions that require summary expressions are the Dimension Functions, except dimcount, filter, rank, squash, and squash_right.
See also: About Spectre Expressions.