Closed HenrikBengtsson closed 5 years ago
In recent R-devel ("R 3.6.0"), we now get:
library("doFuture") registerDoFuture() plan(multisession, workers=2) library("plyr") x <- list(a = 1:10, beta = exp(-3:3), logic = c(TRUE, FALSE, FALSE, TRUE)) y <- llply(x, quantile, probs = (1:3) / 4, .parallel = TRUE) Error in { : task 3 failed - "'probs' outside [0,1]"
This is independent of doFuture, e.g. it occurs also with doParallel, and appear to be a plyr issue. See https://github.com/hadley/plyr/issues/292.
What's concerning is that this bug may pass without producing an error, but where the results of .parallel=FALSE differs from those of .parallel=TRUE, e.g.
.parallel=FALSE
.parallel=TRUE
> y0 <- llply(list(1:10), quantile, probs = (1:3)/4, .parallel = FALSE) > y1 <- llply(list(1:10), quantile, probs = (1:3)/4, .parallel = TRUE) > y0 [[1]] 25% 50% 75% 3.25 5.50 7.75 > y1 [[1]] 325% 550% 775% 3.25 5.50 7.75
Fixed in R-devel r76227. See also https://github.com/hadley/plyr/issues/292#issuecomment-472223530.
In recent R-devel ("R 3.6.0"), we now get:
This is independent of doFuture, e.g. it occurs also with doParallel, and appear to be a plyr issue. See https://github.com/hadley/plyr/issues/292.
What's concerning is that this bug may pass without producing an error, but where the results of
.parallel=FALSE
differs from those of.parallel=TRUE
, e.g.