test-gts.R fails

[jeremyrcoyle]

Seems to be internal to the learner as

gts_fit <- hts::gts(ts(as.matrix(abc[, -5])))
train_hmax <- max(abc$time)
test_hmax <- max(unique(test_data$time))
gts_fpreds <- forecast(gts_fit, h = test_hmax - train_hmax)$bts

also fails with non-conformable arrays. This could be due to a recent change in how R handles matrix multiplication. I will disable the test for now, but we should decide if we need to remove the learner.

[jeremyrcoyle]

this also applies to hts

[MichaelChirico]

Part of the issue with not keeping the package on CRAN is these types of issues can crop up. Further, the package lacks any real release system so it's hard to tell what the authors consider "stable" commits. All of the recent commits have CI failures.

[MichaelChirico]

FWIW, I noticed this while updating Matrix from 1.5-4.1 to 1.7.3.

(this also necessitated updates to RcppEigen lme4, hits, and MatrixModels but Matrix itself has been the source of all the other issues so far)