Closed thibautjombart closed 4 years ago
Thank you for jumping on this quickly, Teebz :)
It does fix the problem stated, but creates the problem that the model fitting no longer works fits curves on the edges of the bars instead of the middle. That being said, I don't think I would have come up with anything different given the ggplot2 bug.
A side note: there is also a new fun issue from R 4.0's default of stringsAsFactors = FALSE
: https://travis-ci.org/reconhub/incidence/jobs/660131096#L8198-L8202
This popped up because we never enabled cron jobs on travis for this repo, but live and learn.
I have recompiled the doc with R 3.6.3 and devtools_2.2.2 but did not find any immediate fix for the warnings on fit.Rd
. I cannot reproduce the issues reported on Travis.
I cannot reproduce the issues reported on Travis.
This is because it only exists on R-devel. The solution is to explicitly state stringsAsFactors = TRUE wherever data frames are being created (likely in the tests files).
I have not chased the source of the warning, working flat out on covid19 work. Will call for help on RECON slack to finish the PR. More and more people are reporting this issue.
@zkamvar @thibautjombart seems that fixed the isses in the tests
This is a patch for
plot.incidence
which is currently broken due to this issue in ggplot2, as described in https://github.com/reconhub/incidence/issues/119.Outline of patch
geom_col
replacesgeom_bar(..., stat = "indentity"
)This fixes the issue on:
Unrelated stuff that came up
I get unrelated warnings re documentation (see below) which I ignored as probably unrelated to this patch.