Different behaviour calling tar_* functions from Rmd

[boshek]

I am working in a windows 10 environment with CRAN versions of targets (0.2.0) and tarchetypes (0.1.0).

When I run something like:

tar_read(some_target)

I get different behaviour depending where I called it. If I run this on the console or from a .R script it runs fine and returns what I am expecting.

If, however, I call this from an .Rmd I get the following error:

Error: utility functions like tar_read() and tar_progress() require a _targets/ data store produced by tar_make() or similar. 

So far deleting the _targets directory and re-running as well as a fresh clone of the same code both result in this same differential behaviour.

Calling targets:::path_store from both environments returns "_targets". For whatever reason, when called from an .Rmd this line returns TRUE.

Anyone have any ideas what might be happening here?

[wlandau]

Sounds like your R Markdown document is running from a different root directory from where you run tar_make(). Ideally they should agree. Or, you can run something like withr::with_dir("/my/targets/project/root/dir/", tar_read(my_target)) when you want to read a target in an Rmd. Alternatively, you can set knitr::opts_knit$set(root.dir = "/my/targets/project/root/dir/") in a code chunk at the top of the document.

[boshek]

Ah! You are totally right. But what happened was some of the RStudio Rmarkdown settings got monkeyed with behind the scenes. Thanks!