The documentation (e.g., p. 55 in the current https://cran.r-project.org/web/packages/SuperLearner/SuperLearner.pdf ; the Examples section of the help for SuperLearner) says
snow example with user-generated wrappers
If you write your own wrappers and are using snowSuperLearner()
These new wrappers need to be added to the SuperLearner namespace and exported to the clusters
Using a simple example here, but can define any new SuperLearner wrapper
my.SL.wrapper <- function(...) SL.glm(...)
assign function into SuperLearner namespace
environment(my.SL.wrapper) <-asNamespace("SuperLearner")
However, as far as I can tell, adding the wrappers to the namespace is not necessary (exporting them to the cluster is necessary).
I did not read this thoroughly the first time, and I've been using SuperLearner with snow clusters and my custom wrappers without adding them to the namespace; I actually do not understand why it would be necessary to add them to the namespace so I am hesitant to do it. What am I missing?
The documentation (e.g., p. 55 in the current https://cran.r-project.org/web/packages/SuperLearner/SuperLearner.pdf ; the Examples section of the help for SuperLearner) says
However, as far as I can tell, adding the wrappers to the namespace is not necessary (exporting them to the cluster is necessary).
I did not read this thoroughly the first time, and I've been using SuperLearner with snow clusters and my custom wrappers without adding them to the namespace; I actually do not understand why it would be necessary to add them to the namespace so I am hesitant to do it. What am I missing?