Re: [R-pkg-devel] Package with JS dependency

[Michael Chirico]

Here are 2,800 JavaScript files in CRAN packages:

https://github.com/search?q=org%3Acran+path%3Amin.js+-path%3ALICENSE.txt&type=code

You might have a look at those. To Duncan's point, note there are also
0 min.js files:

https://github.com/search?q=org%3Acran+path%3Amin.js+-path%3ALICENSE.txt&type=code

Mike C

On Fri, Feb 21, 2025 at 9:01 AM Duncan Murdoch  wrote:
>
> On 2025-02-20 12:36 p.m., Brian Leonard wrote:
> > Hello there,
> >
> > [TLDR: I'm curious to know the degree to which including a pure javascript 
> > program (via the V8 package) in an R package affects the potential of 
> > passing CRAN review]
> >
> > I'm working on implementing an R package to support users of the Psych-DS 
> > data specification. The package will include a 
> > shiny app with helpful functions for assembling and publishing high-quality 
> > datasets in the behavioral sciences, but one of the core functions will be 
> > to validate existing datasets (local directories) according to the rules of 
> > our schema.
> >
> > We already have this validator functionality set up as a Deno application, 
> > and we can use certain Deno tools, along with esbuild, to generate 
> > node-based and pure-javascript bundled versions of validator.
> >
> > We'd like to be able to leverage our existing code instead of having to 
> > reinvent the wheel for the R context. If we try to publish our package to 
> > CRAN with this javascript script as a core part of the package's 
> > functionality, will that severely affect our chances of passing the review 
> > process? I know that having sufficient unit tests and well documented 
> > processes is crucial to getting a package through, but the validator code 
> > in this scenario would be stringy and minified, and the bundle would not 
> > include unit tests for all the validator's internal functions. Is it the 
> > case that a dependency like this is equally subject to CRAN review 
> > scrutiny, or would it be sufficient to have all the R-based components 
> > comply with CRAN requirements?
>
> I am not a member of CRAN, so I can't give you authoritative advice on
> this.  But I think one issue is that if you are only including minified
> code, then they wouldn't see that as the original source.  My rgl
> package includes a few thousand lines of Javascript code which is
> minified when installed; the tarball contains the original source.
>
> Duncan Murdoch
>
> __
> R-package-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel

__
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel

Re: [R-pkg-devel] Package with JS dependency

[Duncan Murdoch]

On 2025-02-20 12:36 p.m., Brian Leonard wrote:

Hello there,

[TLDR: I'm curious to know the degree to which including a pure javascript 
program (via the V8 package) in an R package affects the potential of passing 
CRAN review]

I'm working on implementing an R package to support users of the Psych-DS data 
specification. The package will include a shiny 
app with helpful functions for assembling and publishing high-quality datasets in the 
behavioral sciences, but one of the core functions will be to validate existing 
datasets (local directories) according to the rules of our schema.

We already have this validator functionality set up as a Deno application, and 
we can use certain Deno tools, along with esbuild, to generate node-based and 
pure-javascript bundled versions of validator.

We'd like to be able to leverage our existing code instead of having to 
reinvent the wheel for the R context. If we try to publish our package to CRAN 
with this javascript script as a core part of the package's functionality, will 
that severely affect our chances of passing the review process? I know that 
having sufficient unit tests and well documented processes is crucial to 
getting a package through, but the validator code in this scenario would be 
stringy and minified, and the bundle would not include unit tests for all the 
validator's internal functions. Is it the case that a dependency like this is 
equally subject to CRAN review scrutiny, or would it be sufficient to have all 
the R-based components comply with CRAN requirements?


I am not a member of CRAN, so I can't give you authoritative advice on 
this.  But I think one issue is that if you are only including minified 
code, then they wouldn't see that as the original source.  My rgl 
package includes a few thousand lines of Javascript code which is 
minified when installed; the tarball contains the original source.


Duncan Murdoch

__
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel

[R-pkg-devel] Package with JS dependency

[Brian Leonard]

Hello there,

[TLDR: I'm curious to know the degree to which including a pure javascript 
program (via the V8 package) in an R package affects the potential of passing 
CRAN review]

I'm working on implementing an R package to support users of the Psych-DS data 
specification. The package will include a shiny 
app with helpful functions for assembling and publishing high-quality datasets 
in the behavioral sciences, but one of the core functions will be to validate 
existing datasets (local directories) according to the rules of our schema.

We already have this validator functionality set up as a Deno application, and 
we can use certain Deno tools, along with esbuild, to generate node-based and 
pure-javascript bundled versions of validator.

We'd like to be able to leverage our existing code instead of having to 
reinvent the wheel for the R context. If we try to publish our package to CRAN 
with this javascript script as a core part of the package's functionality, will 
that severely affect our chances of passing the review process? I know that 
having sufficient unit tests and well documented processes is crucial to 
getting a package through, but the validator code in this scenario would be 
stringy and minified, and the bundle would not include unit tests for all the 
validator's internal functions. Is it the case that a dependency like this is 
equally subject to CRAN review scrutiny, or would it be sufficient to have all 
the R-based components comply with CRAN requirements?

Thanks,
Brian Leonard
Software Engineer, Psych-DS (MIT)



[[alternative HTML version deleted]]

__
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel