[R-pkg-team] Bug#1123933: Bug#1123933: r-cran-glmmtmb: autopkgtest regression with r-cran-reformulas 0.4.3-1

[Dirk Eddelbuettel]

Hi Charles (and Hi Graham),

On 26 December 2025 at 21:11, Graham Inggs wrote:
| Hi Charles
| 
| On Thu, 25 Dec 2025 at 00:06, Charles Plessy <plessy at debian.org> wrote:
| > But I would like to ask: why is the bug marked regression?  The tests
| > are failing in all suites including Stable since the release, so I do not
| > see the relation with r-cran-reformula's update yesterday?

As I am not the maintainer of r-cran-glmmtmb, I did not get your reply in
this thread. Thanks to Graham for adding me in CC.

Without having looked into details I suspect this has little to do with the
update of r-cran-reformulas, but likely a bit more with the fact that
r-cran-glmmtmb is three releases behind.

Now, (r-cran-)reformulas is a fairly recent carveout. When that happens it
sometimes takes a release or two to shake out issues like the one quoted in
this thread (on `[[2]]` subset to a list). This may have driven parts of the
three updates in (r-cran-)glmmtmb that we are now behind. 

| r-cran-glmmtmb's autopkgtests were passing in March 2025.
| 
| The four failing tests; "basic smooth", "smooth with no fixed-effect",
| "smooth + diag() specials" and "multiple smooths" only started to fail
| after the update of r-cran-reformulas.  I checked again, and the
| failures first appeared on 2025-11-01, after the upload of version
| 0.4.2-1, not version 0.4.3-1 as I wrote previously.

Sounds entirely plausible.


While I have you both here:  r-cran-bit64 still needs help or we get a big
autoremoval wave on Jan 11.  Charles you Dec 12 upload helped but did not
unblock this:  https://tracker.debian.org/pkg/r-cran-bit64

Do we need to remove the i386 / armhf / s390x binaries from the repo to
unblock this?

Best, Dirk
| 
| Regards
| Graham

-- 
dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org