[FR] preserve spacing between source code lines in a code block

[IndrajeetPatil]

Description of the issue

When all the source and output blocks from one code chunk are collapsed into a single block, any spaces that were deliberately left to separate different code lines are not preserved in the rendered document.

Source code

```{r, collapse=TRUE}
x <- 1
y <- 2

# addition
x + y

# subtraction
x - y

# multiplication
x * y
```

Current output

x <- 1
y <- 2

# addition
x + y
#> [1] 3
# subtraction
x - y
#> [1] -1
# multiplication
x * y
#> [1] 2

Expected output

x <- 1
y <- 2

# addition
x + y
#> [1] 3

# subtraction
x - y
#> [1] -1

# multiplication
x * y
#> [1] 2

[yihui]

You can use the chunk option strip.white = FALSE (https://yihui.org/knitr/options/). Next time you may ask this type of question on Stack Overflow or RStudio Community. Thanks!

[IndrajeetPatil]

Apologies for violating the issue guidelines.

I am surprised I missed the existence of this chunk option when I've read that webpage so many times by now. 😓

[yihui]

No worries! That's very normal. I miss information under my nose sometimes, too.

[cderv]

I think you may have missed it also because this is a side effect of a change in knitr 1.34 which was just released to CRAN https://github.com/yihui/knitr/releases/tag/v1.34

The chunk option strip.white = TRUE used to work only when the chunk option collapse = FALSE. Now the two options are independent, i.e., strip.white also works when collapse = TRUE (thanks, @kbvernon, #2011).

• strip.white = TRUE is the default in knitr
• Previously, it had no impact on chunk with collapse = TRUE. This means chunks where not stripped before collapsing.
• Now it has - which I believe create the new behavior that requires strip.white = FALSE if you don't want it.

This could probably be qualified as a breaking change and we should probably have communicated this way in our NEWS file to indicated how to put back previous behavior. 😅

Example below:

Using version 1.33

# renv::install("[email protected]")
packageVersion("knitr")
#> [1] '1.33'
content <- c(
  "```{r, collapse=TRUE}",
  "x <- 1", 
  "y <- 2", 
  "", 
  "# addition", 
  "x + y",
  "",
  "# subtraction",
  "x - y"
)
xfun::raw_string(content)
#> ```{r, collapse=TRUE}
#> x <- 1
#> y <- 2
#> 
#> # addition
#> x + y
#> 
#> # subtraction
#> x - y
res <- knitr::knit(text = content)
xfun::raw_string(res)
#> 
#> ```r
#> x <- 1
#> y <- 2
#> 
#> # addition
#> x + y
#> #> [1] 3
#> 
#> # subtraction
#> x - y
#> #> [1] -1
#> ```

Using 1.34

# renv::install("knitr")
packageVersion("knitr")
#> [1] '1.34'
content <- c(
  "```{r, collapse=TRUE}",
  "x <- 1", 
  "y <- 2", 
  "", 
  "# addition", 
  "x + y",
  "",
  "# subtraction",
  "x - y"
)
xfun::raw_string(content)
#> ```{r, collapse=TRUE}
#> x <- 1
#> y <- 2
#> 
#> # addition
#> x + y
#> 
#> # subtraction
#> x - y
res <- knitr::knit(text = content)
xfun::raw_string(res)
#> 
#> ```r
#> x <- 1
#> y <- 2
#> 
#> # addition
#> x + y
#> #> [1] 3
#> # subtraction
#> x - y
#> #> [1] -1
#> ```

[IndrajeetPatil]

Yes, please, let's go with this explanation for why I forgot about that option 😅

[yihui]

I suspected in yihui/knitr#2011 that this might be a breaking change, but I didn't realize a breaking case.

At this point, I feel strip.white = FALSE may be a better default for the case of collapse = TRUE.

[cderv]

Yes I missed it too. I also think strip.white = FALSE may be a better default for collapse = TRUE

[cderv]

I tried to tackle this but did not find exactly where to apply a modification.
Do we have a way to detect if an option is currently default value or set on a chunk by the user ? Because unless we revert, I think here we want to change the default strip.white value when collapse = TRUE on a chunk or a globally, unless specify by the user in which case we want to keep stip.white = FALSE.

Is this done by a knitr hook ? Is there another mechanism ?

I created a knitr issue to follow this change

[cderv]

@IndrajeetPatil we just merged in knitr a change to fix this regression. Thanks for the report!

[IndrajeetPatil]

Wonderful! Glad that the issue was helpful. :)

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue by following the issue guide (https://yihui.org/issue/), and link to this old issue if necessary.