Skip to content

[Bioc-devel] WARNING: Add non-empty \value sections to the following

3 messages · Witold E Wolski, Daniel, Dimitrov, Martin Morgan

#
Hello,

BiocCheck produces the following warning for function with no return value.

WARNING: Add non-empty \value sections to the following man
  pages: x.Rd, y.Rd, z.Rd

However, the functions x, y, and z do not return a value since these are 
functions which, e.g., plot to grDevices, or write to a file.
I am using roxygen2 to create the documentation.

I tried to address the problem by  specifying

#' @return NULL

But roxygen2 does not produce the \value section if @return NULL.

What is the best practice to suppress this warning?

Best,
Witek
#
Dear BioConductor developers,

Two windows users of a package that I'm developing (soon to be submitted 
to BioC) have encountered the same error 
<https://github.com/saezlab/liana/issues/25>:

```
Error: BiocParallel errors
 ? element index: 1, 2, 3
 ? first error: invalid length of row names
```

This error is not reproducible in either Mac, nor Linux, and from my 
search I could pin point it to differences in which variables from the 
parent environment are handled on Windows.

The package that seems to be throwing this error is *scuttle*, and in 
particular the summarizeAssayByGroup function's call to BiocParallel.

Any help would be appreciated.

Kindest regards,

Daniel Dimitrov
#
Linux and macOS use MulticoreParam() by default, and the 'workers' inherit the environment (e.g., loaded packages) of the 'manager'. Windows doesn't support this mode, and instead uses SnowParam() by default; likely you'll be able to reproduce the error on your system by using SnowParam(), perhaps via register(SnowParam()) or explicitly in your function.

Likely, your function assumes that it has access to a variable defined in the global environment; this should be made explicit by passing that variable as an argument. Here's an illustration of the principle (not exactly your error)
[[1]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"

[[2]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"

[[3]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"
Error: BiocParallel errors
  2 remote errors, element index: 1, 2
  1 unevaluated and other errors
  first remote error:
Error in rownames(m): object 'm' not found

and the solution (by passing 'm' explicitly to the function)
[[1]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"

[[2]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"

[[3]]
 [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
[20] "t" "u" "v" "w" "x" "y" "z"

Martin



?On 11/22/21, 12:15 PM, "Bioc-devel on behalf of Daniel, Dimitrov" <bioc-devel-bounces at r-project.org on behalf of daniel.dimitrov at uni-heidelberg.de> wrote:

    Dear BioConductor developers,

    Two windows users of a package that I'm developing (soon to be submitted 
    to BioC) have encountered the same error 
    <https://github.com/saezlab/liana/issues/25>:

    ```
    Error: BiocParallel errors
     ? element index: 1, 2, 3
     ? first error: invalid length of row names
    ```

    This error is not reproducible in either Mac, nor Linux, and from my 
    search I could pin point it to differences in which variables from the 
    parent environment are handled on Windows.

    The package that seems to be throwing this error is *scuttle*, and in 
    particular the summarizeAssayByGroup function's call to BiocParallel.

    Any help would be appreciated.

    Kindest regards,

    Daniel Dimitrov