[Rd] Parametrized Vignettest in R packages

Witold E Wolski wewol@ki @ending from gm@il@com
Mon Jul 9 21:24:57 CEST 2018


Dear Yihui,

Thank you for the valuable questions.

sample_analysis is a "tibble" while
configuration is an "R6" class.
But I also have parametrized reports where I pass R reference classes
as arguments.

This is the Rmd yaml params part corresponding to the error message.

params:
  configuration:  !r get(data(skylineconfig))
  data: !r get(data(sample_analysis))

Might the R6 class which I pass to  configuration the source of the
problem here? I have heard somewhere that R6 uses environments to
implemented some features.



There is also a further problem I am encountering reproducible when
running devtools::install() or R CMD INSTALL

** installing vignettes
   'tr_srm_summary.Rmd' using 'UTF-8'
Warning in data(skylineconfig) : data set 'skylineconfig' not found
Error in get(data(skylineconfig)) : object 'skylineconfig' not found

This problem isn't happening when running
devtools::build_vignettes()

The Package install works however if I first build the vignettes with
devtools::build_vignettes() and than run R CMD INSTALL


I will be sending to you a link to a repository with a reproducible
example in a separate e-mail.

Kind regards
Witek

On 9 July 2018 at 18:24, Yihui Xie <xie using yihui.name> wrote:
> So far you haven't provided a reproducible example yet. I wonder what
> exactly the object `sample_analysis` is. Sounds like it is an
> environment. If that is the case, devtools::build_vignettes() will
> tangle (for the meaning of "tangle", see ?tools::buildVignette) your
> vignette into an invalid R script. Environments cannot be represented
> via dput(), so knitr's purl() (eventually called by the vignette
> engine) cannot really create a valid R script when your vignette
> contains `params` that uses environment values. Most other types of
> values should be fine.
>
> Regards,
> Yihui
> --
> https://yihui.name
>
>
> On Mon, Jul 9, 2018 at 10:37 AM, Duncan Murdoch
> <murdoch.duncan using gmail.com> wrote:
>> On 09/07/2018 8:49 AM, Witold E Wolski wrote:
>>>
>>> Dear Duncan,
>>>
>>> Following your advice (Thank you for it) I did include into the vignettes
>>> params:
>>>    configuration: !r get(data(sample_analysis))
>>>    data: !r get(data(skylineconfig))
>>>
>>> And everything seemed (see below) to work fine.
>>> devtools::build_vignettes() builds the vignettes.
>>>
>>> Runs with NO error. By this the .Rmd file, html file and corresponding
>>> .R files is being placed into the inst/doc directory.
>>>
>>> However, when I did want to check the package by running
>>> devtools::check(build_args="--no-build-vignettes")
>>>
>>> The following warning is shown:
>>>
>>> Warning: parse error in file
>>>
>>> 'C:/Users/wolski/AppData/Local/Temp/Rtmp4QanHv/SRMService.Rcheck/SRMService/doc/tr_srm_summary.R':
>>>
>>> /Users/wolski/AppData/Local/Temp/Rtmp4QanHv/SRMService.Rcheck/SRMService/doc/tr_srm_summary.R:2:22:
>>> unexpected '<'
>>> 1: params <-
>>> 2: list(configuration = <
>>>
>>> This is how the start of the .R file produced from the vignette looks
>>> like.
>>>
>>> params <-
>>> list(configuration = <environment>, data = structure(list(protein_Id =
>>> c("CiRT standards",
>>>
>>> I am worried that this is the end of this road with parameterized
>>> vignettes then? What do you think?
>>
>>
>> Interesting.  The issue is that the value being assigned to configuration is
>> an environment, and whatever code produces those lines in tr_srm_summary.R
>> isn't deparsing it into something parseable.  (This isn't a big surprise:
>> dput and deparse both fail at that.)
>>
>> I'd guess a solution is not to save sample_analysis as an environment, but
>> maybe you didn't know you were doing that since some other structures are
>> implemented as environments.
>>
>> Perhaps you could use
>>
>> configuration: !r as.list(get(data(sample_analysis)))
>>
>> and then when you need to use it, convert it back to the original type?
>>
>> Duncan Murdoch
>>>
>>>
>>> regards
>>> Witek
>>>
>>>
>>>
>>>
>>>
>>> On 2 July 2018 at 18:21, Duncan Murdoch <murdoch.duncan using gmail.com> wrote:
>>>>
>>>> On 02/07/2018 11:22 AM, Witold E Wolski wrote:
>>>>>
>>>>>
>>>>> Hello,
>>>>>
>>>>> Thank you for the questions as well as remaining me of the default
>>>>> parameters in the yaml session.
>>>>> Indeed this seems to be the solution.
>>>>>
>>>>>
>>>>> But how would I assign package data as a default parameter?
>>>>> So originally I thought to render the markdown with :
>>>>>
>>>>> <code>
>>>>> data(sample_analysis)
>>>>> data(skylineconfig)
>>>>> x <- rmarkdown::render("report.Rmd", output_format = "html_document",
>>>>>                         params = list(data = sample_analysis,
>>>>>                                       configuration =
>>>>> skylineconfig),envir = new.env())
>>>>> <code>
>>>>>
>>>>> I do not think I can write:
>>>>>
>>>>> params:
>>>>>     configuration: !r data(sample_analysis)
>>>>>     data: !r data(skylineconfig)
>>>>>
>>>>> since data does not return the package just puts them in the env as a
>>>>> side effect. Is there a different base function to achieve it.
>>>>
>>>>
>>>>
>>>> I think you could use
>>>>
>>>>   params:
>>>>    configuration: !r get(data(sample_analysis))
>>>>    data: !r get(data(skylineconfig))
>>>>
>>>> but if that doesn't work, a longer version is
>>>>
>>>>   params:
>>>>    configuration: !r {data(sample_analysis); sample_analysis}
>>>>    data: !r {data(skylineconfig); skylineconfig}
>>>>
>>>> Duncan Murdoch
>>>>
>>>>
>>>>>
>>>>>
>>>>> Thank you
>>>>> Witek
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On 2 July 2018 at 16:56, Duncan Murdoch <murdoch.duncan using gmail.com>
>>>>> wrote:
>>>>>>
>>>>>>
>>>>>> On 02/07/2018 10:30 AM, Witold E Wolski wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Hello,
>>>>>>>
>>>>>>> I have a package which includes some parameterized r-markdown report
>>>>>>> which I would also like to build as package vignettes.
>>>>>>>
>>>>>>> Is there a way to run the parameterized vignette creation with the
>>>>>>> package build or package check?
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Doesn't the usual method work?  You can specify defaults for parameters
>>>>>> in
>>>>>> the YAML header; I'd expect those to be the parameter values that get
>>>>>> used.
>>>>>>
>>>>>> You can give instructions to your users on how to rebuild the reports
>>>>>> with
>>>>>> different parameters.
>>>>>>
>>>>>> Duncan Murdoch
>>>>>>
>>>>>>>
>>>>>>> Thank you
>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>>
>>
>> ______________________________________________
>> R-devel using r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel



-- 
Witold Eryk Wolski



More information about the R-devel mailing list