Are these good ideas?

[Thomas Passin]

For parameter passing like your #2, I have packaged them into a 
dictionary and passed that around.  It was easy enough, and worked well.

The only potential problem is in documenting the key/value pairs the 
dictionary is supposed to contain.  You had better make sure it's made 
clear somewhere, preferable where it is consumed.

For "global" values, sure, put them in a module and import them as 
needed.  Just make sure that none of your code is going to mutate those 
values, or you will end up with mass confusion.  I suppose you could 
make them be properties that have no setters, if you want to go to that 
trouble.

On 11/14/2022 12:14 PM, Stephen Tucker wrote:
> Hi,
> 
> I have two related issues I'd like comments on.
> 
> Issue 1 - Global Values
> 
> Some years ago, I had a situation where
> (a) I could supply low-level functions that carry out tasks,
> (b) I needed those functions to communicate with each other, but
> (c) I had no access to the module that invoked my functions.
> 
> In order to achieve this, I hit on the idea that I also supply a module
> that I describe as a "global values" module. This module …
> (i) … defines the values that the functions use to communicate with each
> other;
> (ii) … is the subject of an import statement in each of my functions;
> (iii) … initialises its values at the start of each run (since the import
> only carries out an actual import once per session);
> (iv) … acts as a repository for the values thereafter.
> 
> This solution works well.
> 
> Given that I am not particularly concerned about efficiency,
> (1) Is this a reasonable way to achieve this goal?
> (2) Do you know of any better ways?
> 
> Issue 2 - Passed Parameters
> 
> I am now facing another situation where I am wanting to pass 6 or 7
> parameters down through several layers of logic (function A calling
> function B calling ... ) and for results to be passed back.
> 
> Having had the idea described above, I am considering using it again to
> save all the parameter-and-results passing.
> 
> I see nothing wrong with doing that, but I may well be missing something!
> 
> Comments, please!
> 
> Stephen Tucker.