[R] There is a relationship of the Modify-in-place optimisation of a object and the local variable `*tmp*`?

[Duncan Murdoch]

On 07/10/2021 6:46 p.m., Ben Deivide de Oliveira Batista wrote:
> Dear R users,
> 
> When modify-in-place of objects occurs, is there a local variable called
> `*tmp*`, behind the scenes R? Let's look at two examples to understand the
> question.
> 
> Example 1 (R Language Definition)
> --------------------------------------------------
> 
>> x <- 1:10
>> tracemem(x)
> [1] "<000000000798F758>"
>> x[3:5] <- 13:15
> tracemem[0x000000000798f758 -> 0x0000000008207030]:
> 
> The result of this command is as if the following had been executed
> `*tmp*` <- x
> x <- "[<-"(`*tmp*`, 3:5, value=13:15)
> rm(`*tmp*`)
> 
> Conclusion: Here copy-on-modify occurs!
> 
> 
> Example 2 : Modify-in-place
> -----------------------
> 
>> x <- c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
>> tracemem(x)
> [1] "<0000000008BFB818>"
>> x[3:5] <- 13:15Conclusion: Here modify-in-place occurs!
> For example 2, is there a local variable `*tmp*` for this case? If so,
> what would the syntactic representation look like, similar to example 1?

I think there is, and it would be the same as in 1.  The memory 
operations are different because x is different:  in 1, it is stored in 
a compact representation of 1:10, which needs to be expanded to a full 
array of doubles before the entries can be changed.  In 2, it is already 
stored as an array of doubles so this isn't needed.

Duncan Murdoch