wflow_build with local=TRUE doesn't leave objects in global namespace

[JohnReid]

After calling wflow_build('Singh-DeLorean.Rmd', local=TRUE), I expected to have all the objects created in the Rmarkdown file available in my global namespace. This doesn't seem to be the case. Are they somewhere else? What does local=TRUE actually do? I'm not using RStudio, could this be the problem?

/usr/bin/pandoc +RTS -K512m -RTS Singh-DeLorean.utf8.md --to html --from markdown+autolink_bare_uris+ascii_identifiers+tex_math_single_backslash --output Singh-DeLorean.html --smart --email-obfuscation none --standalone --section-divs --table-of-contents --toc-depth 3 --variable toc_float=1 --variable toc_selectors=h1,h2,h3 --variable toc_collapsed=1 --variable toc_smooth_scroll=1 --variable toc_print=1 --template /home/john/lib/R/library/rmarkdown/rmd/h/default.html --no-highlight --variable highlightjs=1 --include-before-body /tmp/Rtmp8FGvZY/rmarkdown-str41fb391d3e6b.html --variable navbar=1 --variable body_padding=51 --variable header_padding=56 --variable 'theme:cosmo' --include-in-header /tmp/Rtmp8FGvZY/rmarkdown-str41fb3f160d34.html --mathjax --variable 'mathjax-url:https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML' --include-after-body include/footer.html

Output created: ../docs/Singh-DeLorean.html
Summary from wflow_build

Settings:
  seed: 12345

The following were built locally in the current R session:

../docs/Singh-DeLorean.html
> dl
Error: object 'dl' not found
> ls()
character(0)

[pcarbo]

@JohnReid I can see why you would be confused by this, and @jdblischak I think we should clarify this point in the documentation. I am also thinking that this is a potentially useful feature for testing R Markdown documents when building one of them at a time (I think should be easy to implement in render_site using the default value for input argument envir?).

Function wflow_build calls render_site from the rmarkdown package with envir = new.env(). So your R Markdown script will be executed from inside a new environment, and will have no effect on the global environment.

@jdblischak will likely have more to add.

[jdblischak]

Thanks for reporting this, @JohnReid. I agree with you that local = TRUE should leave the objects in the global namespace.

And @pcarbo is right, this behavior is because I pass the argument envir = new.env() to rmarkdown::render_site() (line). This is actually a remnant from before I had implemented the option to build the files externally. Setting envir = new.env() helped prevent sharing of variables across files. Since the default is to build the files externally, and local = TRUE is for debugging, it makes the most sense to build the file directly in the global environment.

I'll post to this Issue once this update has been merged into the master branch.

[JohnReid]

Great. Thanks for the quick action.

[jdblischak]

@JohnReid I haven't fixed it yet so I am keeping this Issue open so that other users will see it. Once the change is made and merged into master, I'll close it.

[jdblischak]

@JohnReid This feature has been implemented in v0.7.0. See the release notes for other changes (most noticeably wflow_build() now runs wflow_view() automatically by default). You can upgrade to this latest release by running the following:

devtools::install_github("jdblischak/workflowr", build_vignettes = TRUE)

Please try it out and let me know if this now works as expected.

[JohnReid]

Thanks for the quick response, this works as expected.