Issues with Creating EMARs

[abhsarma]

We want to support export of multiverse analysis to EMARs. While we have a working implementation thanks to @hwangkyle, there are still some pending issues (including #116).

Broadly, this is due to how things are implemented in knitr:

• currently, knitr exports the R engine which can be accessed using knitr::knit_engines$get("R"). But it does not allow us to change the environment in which the code is being evaluated in. This is likely a simple fix, by allowing the environment to be declared as an argument in https://github.com/yihui/knitr/blob/fb1f4231d2df7b156e314686f39fc040ce807513/R/block.R#L187

• the output does not change unless I set plot_counter(reset = FALSE) in the following line: https://github.com/yihui/knitr/blob/fb1f4231d2df7b156e314686f39fc040ce807513/R/block.R#L317
  I am not sure what plot_counter() does or why we need it so I don't know what the solution is here.

[abhsarma]

For the second issue, I think it can be solved by again providing an optional argument to eng_r to control the behavior of plot_counter() which will almost always be FALSE except for this specific use case. But I am apprehensive of just jumping to this solution. @mjskay thoughts?

[mjskay]

hmm I dunno. What is plot_counter()?

[abhsarma]

It's a knitr internal function but it's not documented / described, so I have no idea---which is why I'm asking

[mjskay]

hmm I dunno, try reading or stepping through the code to figure out what it does? Or ask the knitr folks...

[abhsarma]

the second one seems like is solved by changing the chunk label during knitting to be unique

[abhsarma]

This has been filed as an issue / PR in knitr. See: yihui/knitr#2358
I am hoping this will be fixed when they have updated it on their end, if not it will be reopened.

[mjskay]

As a process suggestion, if this issue is not fixed yet here and is waiting on a change in knitr, I would leave it open so that it is still tracked here (and someone coming along with that problem can find it)