As shown in your link to knitr/R/cache.R, knitr sets up a list of functions to manage the cache and stores it in the unexported knitr:::cache object. That object is a list of functions; those functions share as their environment the evaluation frame from when new_cache() was called.
If you wanted to change the location where cached objects were stored, or the method used to store them, you could modify the knitr:::cache object.
This is a little tricky, because it's not an exported object. But here's how to do it:
newcache <- knitr:::cache
# Make all the changes you like to newcache!
assignInNamespace("cache", newcache, "knitr")
For example, this document prints some info about calls to exists in the rendering log:
---
title: "Untitled"
date: "2022-07-29"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, cache = TRUE)
newcache <- knitr:::cache
oldexists <- newcache$exists
newcache$exists <- function(hash, lazy = TRUE) {
cat("cache$exists called with hash=", hash, "\n")
oldexists(hash, lazy)
}
assignInNamespace("cache", newcache, "knitr")
```
```{r}
x <- 1:100
print(x[1:10])
```
I saw this in the log before the usual call to Pandoc:
processing file: Untitled.Rmd
|.................. | 25%
ordinary text without R code
|................................... | 50%
label: setup (with options)
List of 1
$ include: logi FALSE
|.................................................... | 75%
ordinary text without R code
|......................................................................| 100%
label: unnamed-chunk-1
cache$exists called with hash= Untitled_cache/html/unnamed-chunk-1_bc684450204aad4eef880fd842708be8
cache$exists called with hash= Untitled_cache/html/unnamed-chunk-1_bc684450204aad4eef880fd842708be8
