[Bioc-devel] Error in trio package due to removed package

[Ivan Krylov]

В Wed, 1 Apr 2026 12:54:58 +0000
Holger Schwender via Bioc-devel <bioc-devel using r-project.org> пишет:

> Rewriting the function in trio so that it would not depend on the
> haplo.stats function would be hardly possible, in particular
> considering the BioConductor release schedule. Even though this
> function is not one of the main features of the trio package, it is
> part of a workflow. So simply removing (temporarily) this function
> would also be not that simple (and not a really good idea).

Meanwhile, you can also try to contact the haplo.stats maintainer and
tell them that

 * the failure is reproducible with a build of R configured with
   --disable-long-double, Apple hardware is not required
 * the comparison in haplo.stats:::anova.haplo.glmlist fails because
   sum(fit.gaus$haplo.post.info$post) is 220 + 3.410605e-13 instead of
   exactly 220
 * the check in the anova.haplo.glmlist method should include enough of
   a tolerance to accommodate rounding errors, or maybe just call
   round() if the numbers being compared should always be whole

Except for this one problem, the package is in a good shape and is
worth trying a CRAN resubmission.

-- 
Best regards,
Ivan

[1] https://www.gnu.org/licenses/gpl-faq.en.html#AllCompatibility