[Rd] recent changes to r-devel with package date checking?

[Simon Urbanek]

Ben,

thanks for the report!

Dirk,

more below


> On 27/03/2026, at 02:37, Dirk Eddelbuettel <edd using debian.org> wrote:
> 
> 
> On 26 March 2026 at 16:00, Ivan Krylov via R-devel wrote:
>> It turns out that when building a Debian package from an R package, | /usr/share/R/debian/r-cran.mk sets the timestamp based on the Debian | changelog entry: https://bugs.debian.org/774031
>> 
> That is by-now fairly old code related the (very laudable) project of creating 'reproducible builds' [1].  As one easy way to break a reproducible build is to encode the current time (as time tends to vary ...) we took an easy out to stick with 'recent but fixed time' found in the changelog entry.
> 
> If these last-minute-literally-before-4.6.0 changes broke that date processing then that may be a piece we need to pick up.


FWIW the date comparison have been in place for a while and is unrelated to the recent changes, but, yes, it's easy enough to a) treat unrecognized timestamps as simply "old" and possibly b) add support for the Debian timestamps so I'll look into that.

Cheers,
Simon



>  It should however only affect production of r-base-core_*.deb and friends.  Which AFAIK is not done regularly at CRAN. I do rebuild my local R-devel every few days both locally and a schedule Docker build for drd and friends, I do not auto-build the .deb files so I have not tried since the R 4.6.3 builds. It may actually be a good idea to put that task on a schedule at R(-devel) side.  If anybody in R Core wants to do so I'd be happy to assist off-line.
> 
> Dirk
> 
> [1] This was once a within-Debian initiative but it has spread much wider, see https://reproducible-builds.org/ -- and it is also a topic the R community at large does not seem to have picked up on despite all the interest in version pinning and freezing and whatnot because 'repro'.
>