Re: for a given emacs session: insert a subject with an increasing counter

[Emanuel Berg <moasenwood@zoho.eu>]

Eric Abrahamsen wrote:

>> Oh, no!
>>
>> I didn't know of this (never seen it). I have used global
>> variables to hold the "state" (be a memory between function
>> calls) but I've also done more involved solutions like
>> properties and even self-modifying code :O
>>
>> And all the while, it was this easy ...
>
> TBH, I've never actually used a closure in anger, probably
> just because it doesn't occur to me. I'll bet if I went back
> and looked over the code I've written in the past I could
> find some situations where I could have used them, but...
> *shrug*

I think I heard that you shouldn't use, or that you should
avoid, global variables the first time when I was 12.
I understood it enough to stick by it and I guess my code was
good to me 5-6 days out of 7 at least.

However it would have been even better if it had been
formulized like this ...

  When the situation is <something>, while a global variable
  would solve it, <something else> is preferable ...

So, doing now what we were unable to do at age 12, we can
formulate one such rule:

  When the situation is like that, that a function needs
  a state or local memory, which cannot be reset from one
  function call to the other, use a closure!

It is easy to add rules ...

When the value has the nature of an user option, a global
variable is fine ...

But what about when two (or more) functions need access to the
same variable, how should that be done? Local functions
(`cl-labels'), parameters, evaluation - don't know, is there
a short answer? - or can one have several functions in a/one
closure `let'?

Aaanyway, I think the best way to approach this is to grep the
source for `defvar' (or baseline `setq'), prune the options,
and then ...

(Unbelievable that I never read this in a single computer book
of all the tons I've read. Unbelievable people who write
them. Or?)

-- 
underground experts united
https://dataswamp.org/~incal