Parallel reports are dynamic documents where the text and the code are purposefully disentangled.

• Intent. Write exploratory reports that evolve into final reports.
• Scenario. You want to have a completely dynamic report but your development time is impaired or you otherwise have a hard time writing complicated R code in knitr chunks.
• Solution. Write R code in normal .R files and load it into dynamic reports using the knitr::read_chunk function.

1. Interactive development of reports is hard to do right, in a way that encourages you to write the best code with the greatest payoffs down the road.
2. There will always be more code than room on the final report or presentation.
3. It's sometimes hard to reuse code used for literate documentation.

The solution is to write normal .R scripts and use the knitr function read_chunk to load them into my dynamic documents:

```{r}
library(knitr)
read_chunk("report.R")
```

The read_chunk function looks at .R scripts and loads chunk by label looking for special comments like this:

# ---- label-1
# df <- read.csv("...")
# ...
# ---- label-2
# ggplot(df, ...) +
# ...

Then, these chunks can be used in Rmd reports by creating an empty chunk with a label name matching one loaded by read_chunk. These empty chunks take up very little space in the report, yet they allow the report to remain entirely independent. knitr fills in the content of the chunks based on the label when it is compiled.

The point of separating chunk files from the .Rmd report is that sometimes you want to be able to jump in and run all of the code easily, move things around, and refactor at will. To get the most mileage out of this design pattern ensure you can always open up any of the chunk files and run them without issue in an empty environment.

• Parallel reports is basically undoing some of what is great about interleaving documentation and code. Use only if you find it difficult to be productive on long documents.
• There is some inevitable duplication. Keep chunk labels as intuitive as you can.