Do not change order of positive and negative values in PositionStack #3471
Conversation
| @@ -195,7 +195,7 @@ PositionStack <- ggproto("PositionStack", Position, | |||
| ) | |||
| } | |||
|
|
|||
| rbind(neg, pos) | |||
| rbind(neg, pos)[match(seq_len(nrow(data)), c(which(negative), which(!negative))),] | |||
There was a problem hiding this comment.
I don't see immediately that this is always correct. Can there be problems with NAs? Or will NAs never reach this point?
There was a problem hiding this comment.
An NA value in ymax is pretty nonsensical and would also have broken the previous code. I can put in some guarding if you’d like nonetheless
There was a problem hiding this comment.
No strong opinion either way. Just make sure to look into why the CI builds fail.
There was a problem hiding this comment.
The failures are due to the downstream changes in sf, which have been fixed in master but not yet merged into the v3.2.1-rc branch
|
This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/ |
fixes #3390
@karawoo I have checked that this fix does not cause regressions in #1552. Basically what this does is avoid positive and negative y-values to get reordered, not fiddle with the group-based reordering