Persisting variables to global scope (M spec's 'shared' keyword?)

Question

Hi,
I wanted to declare multiple variables I'd like to reuse in a query, and be able to keep them in scope afterwards for use in other queries. The alternative of redeclaring them in each query separately would seem tedious, while individual queries can also
only return one variable each (be it a table/list).
I read in the M formula specification that it should be possible to global-scope a variable using the 'shared' keyword, though apparently this was only for use in the 'sections' part of the M specification unimplemented in Power Query itself. Am I missing
something? Might there be some other way to persist a variable to the shared (global) scope from within a query?

Answer 1

Thank you for the further clarification on sections; I didn't really know they are still kind of there in the background.
Reconsidering, I suppose my perceived issue isn't really so bad. I think what I wanted (using M functions from text files like an external library) is pretty well doable, so I've Github'd (https://github.com/tycho01/pquery) what I have for others to try;
perhaps that may help illustrate my use case a bit. I think I'm actually satisfied now, though I may as well explain better what I was trying.
What I'm imagining (as an analogy to my experience with Excel VBA) is basically having a larger library of user defined M functions to flesh out the built-in ones, with smaller functions for relatively basic tasks, and mid-level functions building upon those
and upon each other.
Considering the (current) effort in adding a bunch of functions to different workbooks, not to mention edit syncing concerns, keeping such all-purpose functions separate from individual workbooks (in analogy to VBA's 'personal' workbook) would seem sensible.
If that were to work out, eventually it may be preferable if things were to 'just work', in terms of all functions being in scope, so each such library function could freely call other such library functions.
At present, if you load multiple functions through Expression.Evaluate(), possible in batch through lists/records as you mentioned, they will all be available in the original scope.
However, with no effective way to pass this around (unless environments?), the loaded functions will not know of each other though. What I do now is to explicitly load used library functions in each function where they are needed. Under this limitation of losing
scope between functions, it might be ideal to have all available functions loaded only once.
Reloading in batch (i.e. load the whole library rather than the parts needed) within every new function scope may become less desirable in terms of unnecessary overhead.
I suppose for the most part my scenario is still largely hypothetical though, so I might as well try and see if the general idea would take off in the PQ community for now. In either case, thank you for your replies.