knitr creates a copy of evaluate::evaluate as the default evaluate hook value

[aronatkins]

The default evaluate hook is defined as:

knitr/R/hooks.R

Lines 9 to 14 in 0d648a6

	.default.hooks = list(
	source = .out.hook, output = .out.hook, warning = .out.hook,
	message = .out.hook, error = .out.hook, plot = .plot.hook,
	inline = .inline.hook, chunk = .out.hook, text = identity,
	evaluate = evaluate::evaluate, document = identity
	)

This means that the default hook is tied to whatever version of evaluate was available at the time knitr was installed. If the version of evaluate changes, the knitr hook is not changed to reflect the new definition of evaluate::evaluate.

This explains errors like:

Quitting from lines 9-10 (helloworld1.Rmd)
Error in evaluate_call(expr, parsed$src[[i]], envir = envir, enclos = enclos,  :
  unused argument (include_timing = include_timing)
Calls: <Anonymous> ... call_block -> block_exec -> in_dir -> evaluate -> evaluate_call

Reported here:
https://stackoverflow.com/questions/40198189/r-markdown-error-in-evaluate-call-when-knitting-any-rmd-file

We have also seen this problem when trying to use knitr inside a packrat cache. In the packrat scenario, we may be using the same version of knitr with multiple versions of evaluate (depending on a particular piece of content).

You probably can avoid this issue with something like:

.default.hooks = list(
  # ...
  evaluate = function(...) evaluate::evaluate(...)
  # ...
)

[aronatkins]

Workaround is to re-register the evaluate hook based on the version of evaluate::evaluate available:

knitr::knit_hooks$set(evaluate = evaluate::evaluate)

[yihui]

Sounds like a reasonable fix. Done. Thanks @aronatkins!

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue by following the issue guide (https://yihui.org/issue/), and link to this old issue if necessary.