I am experiencing a failure in the execution of the mailing list analysis. The failure occurs when a message contains two reference Ids. The stack trace of when the failure occurs is the following: [codeface.R] INFO: Traceback: 0: 1: config.script.run({ positional.args <- list("resdir", "mldir", "listname 2: withCallingHandlers(expr, error = function(e) { if (!interactive()) { 3: dispatch.all(conf, conf$mldir, conf$resdir) 4: analyse.sub.sequences(conf, corp.base, release.intervals, repo.path, resdir 5: mclapply.db(conf, 1:length(iter), function(conf, i) { loginfo(paste("Pro 6: mclapply(X, function(i) { conf <- init.db.global(conf) res.local <- F 7: lapply(X = X, FUN = FUN, ...) 8: FUN(1[[1]], ...) 9: FUN(conf, i) 10: make.forest(corp.sub, do.normalise.bound) 11: threads(corp) 12: get.thread.id(refID, ht) [codeface.R] CRITICAL: wrong arguments for subsetting an environment In the threads(..) function of the tm.plugin.mail library one can see how the reference Id is identified and non-scalar refIDs result it data type problems in subsequently called get.thread.id(..) function. This patch address the case of multiple reference Ids by removing the message from the corpus. It would be better to either chose a single reference Id or appropriately deal with multiple reference Ids. Does anyone if we can arbitrarily choose a reference Id or how to appropriately handle messages with multiple reference Ids? Kind regards, Mitchell