-
Notifications
You must be signed in to change notification settings - Fork 22
Description
I like the concept of Rmd partials for repetitive things I do (like summaries of models, robustness checks, making a codebook for the items in a survey).
But I'm not sure I'm using them right, because it seems a bit hacky at times.
To give an example: here, you can see where I used one to summarise a regression with some plots and headlines (this is a simplified example, it's usually a bit more involved).
The partial itself: https://rubenarslan.github.io/repro_web_stack/_regression_summary.html
Used: https://rubenarslan.github.io/repro_web_stack/3_analyses.html
Here are things that felt a bit hacky:
- I needed to set
options(knitr.duplicate.label = 'allow')
globally if I want named chunks. - I "needed" this convenience function: https://github.com/rubenarslan/formr/blob/master/R/rmarkdown_options.R#L234 so that it's rendered "asis" and so that the environment is the calling environment for knit_child.
- I wanted to call the partials from a function (could simply be
knit_print.lm
or something more "custom" like I use now). So it takes arguments from the calling environment, but I cannot specify these in the Rmd file. I tried using parameters, but I think knitr:knit_child does not support them.
I was thinking of packaging some partials that I've made for myself, but wanted to collect your opinion on best practices first. Should I maybe use an htmlwidget-style approach for this as this is more "standard"? I like about the knit_child approach that I can treat it like text, concatenate several etc.