What is PCAmixdata's recodqual() doing, and what is its failure condition?

Question

I'm trying to debug a call to kmeansvar in some data I'm working on, but before I can even try to reproduce the issue, I have to figure out what this error message from recodqual() is saying:

Error in recodqual(X.quali) : There are columns in X.quali where all the categories are identical

The source is fairly short, but without comments, and I don't understand the test condition or the error message:

recodqual <-
function(X)
    {
        X <- as.matrix(X)
        GNA <- tab.disjonctif.NA(X)
        G <- replace(GNA,is.na(GNA),0)
        ns <- apply(G,2,sum)
        nmiss <- apply((apply(GNA,2,is.na)),2,sum)
        n <- nrow(X)
        if(sum((n-nmiss)==ns)!=0) stop("There are columns in X.quali where all the categories are identical")
        return(G)   
}

What is the stop message trying to say, and what is it testing?

• I thought it was failing because of identical columns, but I got this error with a data.table of only two different categorical variables.
• If made sure I had a data.table with two different categorical variables with more than one unique value per column, the call succeeded.

• However, filtering the categorical subset by uniqueness in rows resulted in the call failing as above:

  cat.cols <- Filter(function(col) { 
    !is.numeric(main.data[,get(col)]) &
      length(main.data[,unique(get(col))]) > 1
  }, colnames(main.data))
  # followed by call using main.data[,cat.cols,with=FALSE]
  

I'll add some test data to reproduce shortly.