[R-pkg-team] Bug#980809: rmatrix: breaks autopkgtest of r-cran-glmmtmb on s390x

[Graham Inggs]

Hi Dirk

On Sun, 31 Jan 2021 at 16:13, Dirk Eddelbuettel <edd at debian.org> wrote:
> This is a bug in glmmTMB.

I don't see how you can be so sure.

> All the 'gradient function must ...' errors are from its tests and
> points to me to a recent change in R (where boolean comparisons can now check
> if the compared vector is corre.  The Matrix change may simply be the trigger.

This seems to be unrelated to the version of R.  If you look at the
test results in Ubuntu [1], the failures started with the upload of
rmatrix/1.3-0-1, continued with rmatrix/1.3-2-1, and then went away
when rmatrix was reverted to 1.2-18.

This doesn't mean there's a bug in rmatrix, it could be an expected
change in behaviour (on big-endian systems).  If this is the case,
then r-cran-glmmtmb needs to be adapted for this new behaviour.

Regards
Graham


[1] https://autopkgtest.ubuntu.com/packages/r/r-cran-glmmtmb/hirsute/s390x