On 19/07/2022 20:32, Ivan Krylov wrote:> On Tue, 19 Jul 2022 16:53:53 +0000
> "Koenker, Roger W" <rkoenker at illinois.edu> wrote:
>
>> I wondered if anyone had a suggestion for an alternative way to
>> reference such things? And incidentally wondered whether DOI links
>> were often this flaky.
>
> Since this discussion does happen on R-package-devel from time to time,
> unfortunately, DOI links can be very flaky, but not for the same
> reasons. For example, see:
> https://stat.ethz.ch/pipermail/r-package-devel/2022q2/008089.html
>
> (In my opinion, this is more of a package development question than an
> R development question and thus would be a better fit for
> <r-package-devel at r-project.org>.)
>
>> I wondered if anyone had a suggestion for an alternative way to
>> reference such things?
>
> I suppose you could only use the ISBN only, but most kinds of URLs you
> could use to link to your book can be expected to stop working sooner
> than the DOI.
>
> A link to a web.archive.org snapshot of the page should last relatively
> long, too. Could Zenodo link to the web page for your book (with a
> different DOI) while CUP figures out their availability problems?
>
>> I was led to believe that DOIs were like Platonic solids always there
>> when you needed them.
>
> There are two parts to DOI checks in R documentation. The one that
> usually fails is `R CMD check` not behaving enough like a human being
> with a browser and tripping anti-robot protection on a journal website.
> This happens so frequently that there's a separate sub-page on URL
> checks linked by the CRAN policy:
> https://cran.r-project.org/web/packages/URL_checks.html
Note that 'R CMD check' does not check URLs -- that is part of the
CRAN-specific checking emulated with --as-cran.
> The one that broke in your case is a publisher failing to keep their
> DOI link working. I knew this wasn't impossible, but have never seen an
> example until now. Unfortunately, I don't know whether CRAN would agree
> to make an exception for a DOI that should be valid for all intents and
> purposes and is only temporarily broken for reasons of your control.
> Maybe they would.
>
The issue is that the DOI is the main information in the Description:
"See Koenker (2006) <doi:10.1017/CBO9780511754098> and Koenker et al.
(2017) <doi:10.1201/9781315120256>"
Roger has said that it refers to a book, but we currently have no way of
knowing from that. The traditional way of citing (giving the book
authors, year, title and publishers) is much less fragile.
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Emeritus Professor of Applied Statistics, University of Oxford