[R-pkg-devel] new maintainer for CRAN package XML

[Henrik Bengtsson]

Some thoughts for what the next steps could be:

I could be wrong, but I doubt that there's someone out there that
wants to pick up 'XML' and start developing it beyond keeping the
lights on.  I also think the community agrees that 'xml2' is the
recommended package to use for XML-related needs.  If this can be
agreed upon, I think the best path forward would be to declare 'XML'
being deprecated. This is some that the current maintainer (CRAN Team)
could already do, but updating the document and the package
Description field to:

[WARNING: The 'XML' package is deprecated. Please do not add it as a
dependency to your R package.] Many approaches for both reading and
creating XML (and HTML) documents (including DTDs), both local and
accessible via HTTP or FTP. Also offers access to an 'XPath'
"interpreter".

That would at least help stopping the influx of new packages being
added that depends on XML, especially since not everyone might be
aware of the state of 'XML' and that 'xml2' exists as an alternative.

This also begs the question: Do we need a way to formally declare a
whole package being deprecated?  Is this something that could be added
to the DESCRIPTION file?  Deprecated: TRUE?  Or something richer like
"Lifecycle: deprecated"?  With such a mechanism in place, 'R CMD
check' could give a NOTE for deprecated dependencies.  That could also
be a block for *new* CRAN submission; existing CRAN package should be
accepted ("grandfathered in").  With a richer {deprecation, ...,
defunct} vocabulary, we could produce WARNINGs and eventually ERRORs
too.


/Henrik

On Thu, Jan 25, 2024 at 9:11 AM Heather Turner <ht using heatherturner.net> wrote:
>
> Re: guidance on how to migrate from XML to xml2, these notes from Daniel Nüst may be helpful: https://gist.github.com/nuest/3ed3b0057713eb4f4d75d11bb62f2d66.
>
> Best wishes,
> Heather
>
> On Wed, Jan 24, 2024, at 3:38 PM, Emmanuel Blondel wrote:
> > if XML is deprecated, then what would be the choice for a package
> > maintainer? Move to xml2 probably at some point I assume
> >
> > I use XML in the R packages I've been developing. For some of them, I
> > started before CRAN started being the maintainer, and before xml2
> > inception. The thing is that XML fulfills requirements, it works and
> > fulfills needs of depending packages that made the choice to use it. For
> > this, it deserves to be maintained in CRAN, without having to enter into
> > comparison exercices with other packages that , as of today, may be
> > better to rely on (with certainly very good reasons).
> >
> > Moving to xml2 (or whatever other package), which although I could agree
> > on the principle, can be costly for packages that use extensively XML.
> > Doing so would mean that we first get the assurance that all XML
> > features are covered elsewhere, and can be migrated smoothly.
> >
> > In any case, please acknowledge that this kind of migration may take
> > time and require resources that vary (or even are missing) depending on
> > the package projects. I doubt having CRAN setting a common deadline for
> > retirement is a good way to foster an efficient maintenance of R
> > packages depending on XML. It would be good to receive guidance how to
> > migrate, while ensuring backward compatibility on our package features.
> >
> > Best
> >
> > Le 24/01/2024 à 15:59, Jeroen Ooms a écrit :
> >> On Mon, Jan 22, 2024 at 3:51 PM Uwe Ligges
> >> <ligges using statistik.tu-dortmund.de> wrote:
> >>> Dear package developers,
> >>>
> >>> the CRAN team (and Professor Ripley in particular) has been the defacto
> >>> maintainer of CRAN package 'XML'.
> >>> Our hope was that maintainers of packages depending on XML will migrate
> >>> to other packages for reading XML structures. This has not happened and
> >>> we still see dozens of strong dependencies on XML.
> >> How is this hope communicated? Many R users assume that XML package is
> >> in great shape and the preferable choice because it is maintained by
> >> the CRAN team and r-core members.
> >>
> >> Perhaps one could follow the precedent from the rgdal retirement, and
> >> set a deadline.
> >>
> >> One way to communicate this effectively would be by introducing a
> >> formal deprecation field in the package description. This could then
> >> be displayed on the XML CRAN html page, and when loading the package
> >> interactively. Other packages that import such a deprecated package
> >> could be given a CMD check warning.
> >>
> >> ______________________________________________
> >> R-package-devel using r-project.org mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-package-devel
> >
> > ______________________________________________
> > R-package-devel using r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-package-devel
>
> ______________________________________________
> R-package-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel