A weird error using adaboost

[vadim-cherepanov]

I created a simple model with one {0,1} dependent variable y (int) and many independent double variables, but when I run

gbm(formula = as.formula("y ~ ."), distribution = "adaboost", data = dt,
    n.trees = 1000, interaction.depth = 1, shrinkage = 0.01, cv.folds = 0, keep.data = FALSE)

I get

Error in is.null(weights) || is.infinite(weights) :
'length = NNN' in coercion to 'logical(1)'

where NNN ~ 1500 is the number of observations. There are no NAs in the data, and both 0s and 1s are present in y.

So, I decided to check an older more complicated model, which worked flawlessly before, but after thinking for a while it gave me the same error with a different NNN ~ 30,000.

Any ideas what may cause this now? Thank you.