Minor test/doc/code refactoring #158
Conversation
|
|
||
| default_output_handler <- new_output_handler() |
There was a problem hiding this comment.
First, knitr uses this internal function name
https://github.com/yihui/knitr/blob/aa34f1b85f6bb899b1e23a72ebd64054afb3c695/R/utils.R#L860-L861
So removing it breaking knitr. I did not review this PR so did not test it. Only tried dev evaluate today.
Moreover, this was added some years ago on purpose for performance reason it seems
78059bb
Commit message
for a large number of expressions, evaluating new_output_handler() again and again can take some time and also unnecessary, so just create a default copy beforehand and use it
How should we proceed regarding knitr ? Can we revert this so that knitr is not impacted ? Or should we make knitr do differently ?
But this will mean that older knitr won't work with newer evaluate. Which is already the case IMO as I needed already to do two commits in knitr to prepare for new evaluate (yihui/knitr@5510d3f and yihui/knitr@f4d670f)
There was a problem hiding this comment.
If knitr uses it, I can bring it back (but compute it in a safer way)
No description provided.