Feedback on using the package

[llrs-roche]

Hi! Nice package.

As I was working on something similar I came to check this package. It works well in some cases and provides functionality like the traceability matrix that is not as easily provided by similar solutions. However, I have some comments about improvements it could have (not sure if it is worth it given how the landscape is changing on this area but I wanted to share some feedback):

• It doesn't allow to pick packages from Bioconductor, or from github.

Bioconductor

GOSemSim <- risk.assessr::assess_pkg_r_package("GOSemSim")
 ## Trying https://packagemanager.posit.co/cran/latest
 ## Trying https://cloud.r-project.org
 ## Error: Failed to download the package using remotes::download_version. Error: couldn't find package 'GOSemSim'

• If a specific version is requested the dependencies for that version are not pulled (and installed):

Old knitr version 0.1

risk.assessr::assess_pkg_r_package("knitr", version="0.1")
## Trying https://packagemanager.posit.co/cran/latest
## unpacking C:\Users\sanchosl\AppData\Local\Temp\RtmpkPinDd\file5ca0466e605a.tar.gz locally
## unpacked C:\Users\sanchosl\AppData\Local\Temp\RtmpkPinDd\file5ca0466e605a.tar.gz locally
## installing knitr locally
## knitr is already installed
## Some exported functions are missing help files in knitr
## knitr has bug reports URL
## knitr has a license
## knitr has a maintainer
## knitr has a website
## knitr has a source control
## knitr has no vignettes
## knitr has examples
## knitr has no news
## knitr has no news path
## knitr has no current news
## running code coverage for knitr
## code coverage for knitr unsuccessful
## R coverage for knitr failed. Read in the covr output to see what went wrong: 
##   traceability matrix for knitr unsuccessful
## running rcmdcheck for knitr
## ── R CMD build ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## ✔  checking for file 'C:\Users\sanchosl\AppData\Local\Temp\RtmpkPinDd\temp_file_5ca0298d5941\knitr/DESCRIPTION' (362ms)
## ─  preparing 'knitr':
##   ✔  checking DESCRIPTION meta-information ... 
## ─  checking for LF line-endings in source and make files and shell scripts (518ms)
## ─  checking for empty or unneeded directories
## ─  building 'knitr_0.1.tar.gz'
## 
## ── R CMD check ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
## ─  using log directory 'C:/Users/sanchosl/AppData/Local/Temp/RtmpkPinDd/file5ca06d3a31ba/knitr.Rcheck' (955ms)
## ─  using R version 4.4.2 (2024-10-31 ucrt)
## ─  using platform: x86_64-w64-mingw32
## ─  R was compiled by
## gcc.exe (GCC) 13.3.0
## GNU Fortran (GCC) 13.3.0
## ─  running under: Windows 11 x64 (build 22631)
## ─  using session charset: UTF-8
## ─  using option '--no-manual'
## ✔  checking for file 'knitr/DESCRIPTION'
## ─  checking extension type ... Package
## ─  this is package 'knitr' version '0.1'
## ─  package encoding: UTF-8
## ✔  checking package namespace information ...
## E  checking package dependencies (3.8s)
## Package required but not available: 'highlight'
## 
## Packages suggested but not available for checking: 'rgl', 'tikzDevice'
## 
## See section 'The DESCRIPTION file' in the 'Writing R Extensions'
## manual.

Notably as the package is already installed (no specific library is set it could mess with one's library if they installed an old version of a package) it doesn't have the --keep-source option to run code coverage.

• I haven't checked deep in the code but the traceability matrix doesn't seem to work well. I tested on one package of mine, where I know that some functions are covered by tests and there were some NA values on coverage_percent . They represent almost 44% of the functions when the results$covr_list$total_cov is 0.826!
• The results$dependencies is in a string. It would be useful to have a table and know which ones are on each field.
• It is unclear why the modify_description_file() function is needed (and it doesn't produce any log on the screen).

Thanks for sharing this open source, I might use the traceability matrix.

[Edward-Gillian]

@llrs-roche

Thank you for the comments!

It doesn't allow to pick packages from Bioconductor, or from github.

We are working on functionality to process packages from Bioconductor for the next release!

If a specific version is requested the dependencies for that version are not pulled (and installed):

This is something we will look into!

I haven't checked deep in the code but the traceability matrix doesn't seem to work well. I tested on one package of mine, where I know that some functions are covered by tests and there were some NA values on coverage_percent . They represent almost 44% of the functions when the results$covr_list$total_cov is 0.826!

Can you share the package you were testing the traceability matrix?

The results$dependencies is in a string. It would be useful to have a table and know which ones are on each field.

We are aware of this issue and will be working on it.

It is unclear why the modify_description_file() function is needed (and it doesn't produce any log on the screen).

This function helps resolve this issue when running risk.assessr::assess_pkg_r_package, assess_pkg, and risk_assess_pkg r-lib/pkgbuild#128. In summary, pkgbuild is not dealing with the inst/doc folder correctly and devtools::check() also uses pkgbuild to build a package and thus deletes inst/doc, as well. We will document this issue in the function modify_description_file.

[llrs-roche]

We are working on functionality to process packages from Bioconductor for the next release!

Good to know.

If a specific version is requested the dependencies for that version are not pulled (and installed):

This is something we will look into!

Good, I think this could be useful beyond assessment on pharma environments. If you have challenges let me know, this is something I have some experience and I've dedicated some time to explore.

I haven't checked deep in the code but the traceability matrix doesn't seem to work well. I tested on one package of mine, where I know that some functions are covered by tests and there were some NA values on coverage_percent . They represent almost 44% of the functions when the results$covr_list$total_cov is 0.826!

Can you share the package you were testing the traceability matrix?

Yes, I was testing it with BaseSet. But I think it might be related to the S4 classes not being fully traceable with tracecovr (not sure where I read that).

The results$dependencies is in a string. It would be useful to have a table and know which ones are on each field.

We are aware of this issue and will be working on it.

Great! Note that keeping the version requirements would be also good.

It is unclear why the modify_description_file() function is needed (and it doesn't produce any log on the screen).

This function helps resolve this issue when running risk.assessr::assess_pkg_r_package, assess_pkg, and risk_assess_pkg r-lib/pkgbuild#128. In summary, pkgbuild is not dealing with the inst/doc folder correctly and devtools::check() also uses pkgbuild to build a package and thus deletes inst/doc, as well. We will document this issue in the function modify_description_file.

Thanks for the explanation! This resolves the mystery

[Edward-Gillian]

@llrs-roche

Yes, I was testing it with BaseSet. But I think it might be related to the S4 classes not being fully traceable with tracecovr (not sure where I read that).

Thanks for the tip. The test coverage for S4 functions seems to be working:

The traceability matrix is not reporting this correctly as it is not processing the S4 classes correctly.