>you have to go back and find the body of the loop
If the loop is long enough that you don't naturally remember how it was introduced, that's the problem. In the given example, the use of `p.Age` is literally on the next line of code after ` for _, p := range people`.
> I've never understood what is gained by using `p` instead of spelling it out as `person`.
Wisdom I received from, IIRC, the Perl documentation decades ago: tightly-scoped names should be shorter and less attention-grabbing than more broadly-scoped ones, because you should really notice when you're using a global, and you don't want to suffer attention fatigue. (I'm sure the exact wording was quite different.)
Also because it's better for information density. As I recall, Larry Wall also had the idea that more commonly used language keywords should be shorter than rare ones. Good code uses the locals much more often than globals, so you shouldn't need to expend the same amount of effort on them. (The limiting case of this is functional programming idioms where you can eliminate the variable name completely, in cases like (Python examples) `lambda x: int(x)` -> `int`, or `(foo(x) for x in xs)` -> `map(foo, xs)`.
The problem is that many times I have not read the definition to remember. Debugger puts me into a context where I have to figure out what `p` stands for. I go up the call stack and now there's `s` to be deciphered. Worse is the reuse of `p` for person, product, part, etc. in different contexts.
Debugging is not the only problem. Code is read rarely linearly. Many times I browse different uses of a function, or see how a data structure is modified in different contexts. Looking up single letter variables is just a waste of time.
This is a decent reason, but your original comment is dangerously close to a completely strawman argument of "the argument of the func AverageAge(people []Person) should be called peopleToCalculateTheAverageOfTheirAges because what if I forget the method's name or its purpose while I am reading its text?"
Which, now that I think of it, kinda applies to your argument as well. Yeah, the debugger dropped you into the middle of a function you see for the first time in your life. No matter how you dice it or slice it, this means you'll have to study this function; I personally find that going to the function's beginning and skimming it until the breakpoint is the most reliable (and therefore, on average, fastest) method.
I don't deny the trade-off between variable name length and other factors. But, between 1 and 38, I doubt that the optimal length is at 1. Adding 5 more letters doesn't take anything away, IMHO, while increasing comprehensibility significantly.
> Debugger puts me into a context where I have to figure out what `p` stands for.
`p` stands for "the process in question".
I like to think of single-character vars as idea or topic headers that track the single thing I'm currently up to. I'm rarely working with more than one at a time, frequently it's the only variable, and there are contexts where I wouldn't use them at all.
IMHO if you're in a situation where `p` isn't obvious to you, "something has gone wrong".
If the loop is long enough that you don't naturally remember how it was introduced, that's the problem. In the given example, the use of `p.Age` is literally on the next line of code after ` for _, p := range people`.
> I've never understood what is gained by using `p` instead of spelling it out as `person`.
Wisdom I received from, IIRC, the Perl documentation decades ago: tightly-scoped names should be shorter and less attention-grabbing than more broadly-scoped ones, because you should really notice when you're using a global, and you don't want to suffer attention fatigue. (I'm sure the exact wording was quite different.)
Also because it's better for information density. As I recall, Larry Wall also had the idea that more commonly used language keywords should be shorter than rare ones. Good code uses the locals much more often than globals, so you shouldn't need to expend the same amount of effort on them. (The limiting case of this is functional programming idioms where you can eliminate the variable name completely, in cases like (Python examples) `lambda x: int(x)` -> `int`, or `(foo(x) for x in xs)` -> `map(foo, xs)`.