## We want to identify any data that is in OA that is not in the GL
    DT.split_percs
    DT.mgmtr_summary_tall

    ## Some stores are apeparing in OA BEFORE they appear in the GL 
    ## Here is an example of something causing an error.  
    DT.acc[store_name == "Boomkat"]
    DT.mgmtr_summary_tall[store_name == "Boomkat"]
    DT.compare[store_name == "Boomkat"]

    matchKey(DT.acc, DT.mgmtr_summary_tall, kCols.main)
    DT.compare <- merge(  DT.acc                [               , list(gross.oa=sum(gross, na.rm=TRUE), inOA=TRUE), keyby=kCols.main]
                        , DT.mgmtr_summary_tall [!is.na(storeid), list(gross.gl=sum(gross, na.rm=TRUE), inGL=TRUE), keyby=kCols.main]
                        , all=TRUE
                        )

    # matchKey(DT.split_percs, DT.mgmtr_summary_tall, kCols.main)
    # DT.compare <- merge(  DT.split_percs        [               , list(gross.gl=sum(OA_gross_forDSOM, na.rm=TRUE), inGL=TRUE), keyby=kCols.main]
    #                     , DT.mgmtr_summary_tall [!is.na(storeid), list(gross.oa=sum(gross, na.rm=TRUE), inOA=TRUE), keyby=kCols.main]
    #                     , all=TRUE
    #                     )

    ## If these are NA, then they should be FALSE since not in the respective data
    DT.compare[is.na(inOA), inOA := FALSE]
    DT.compare[is.na(inGL), inGL := FALSE]

    ## Drop any stores that NEVER appear in OA.  These have not been properly matched yet
    setkeyIfNot(DT.compare, kCols.store)
    DT.compare <- DT.compare[!DT.compare[  , all(!inGL), keyby=kCols.store][(V1)]]


    ## clear 
    suppressWarnings( {
        colsToClear <- c("first_date_store_is_in_GL", "last_date_store_is_in_GL", "times_missing_from_GL", "Number_of_times_store_appears_in_GL_for_2013_2014", "Number_of_times_store_appears_in_OA_for_2013_2014")
        DT.compare[, (colsToClear) := NULL]
    })

    ##  Grab the min / max date by store for GL
    suppressWarnings( {
        DT.compare[, first_date_store_is_in_GL := min(date[inGL], na.rm=TRUE), by=kCols.store]
        DT.compare[,  last_date_store_is_in_GL := max(date[inGL], na.rm=TRUE), by=kCols.store]
    })


    if (!any(DT.compare[, inOA & !inGL & (gross.oa != 0 | is.na(gross.oa))]))
        warning ("There are no offenders that in OA but not in GL")


    DT.compare[(date >= "2013-01-01" & (gross.gl != 0 | is.na(gross.gl))), Number_of_times_store_appears_in_GL_for_2013_2014 := lunique(date[inGL]), by=kCols.store]
    DT.compare[(date >= "2013-01-01" & (gross.oa != 0 | is.na(gross.oa))), Number_of_times_store_appears_in_OA_for_2013_2014 := lunique(date[inOA]), by=kCols.store]
    ## For those rows in OA but not in GL,  consider them "missing" if the date is strictly between the min and max
    ##  also, make sure gross is none-zero
    DT.compare[(inOA & !inGL & (gross.oa != 0 | is.na(gross.oa))), times_missing_from_GL := {lunique(date[(date > first_date_store_is_in_GL) & (date < last_date_store_is_in_GL)])}, by=kCols.store]
    stopifnot(!is.null(DT.compare[["times_missing_from_GL"]]))

    if (!any(DT.compare[, removeNA(times_missing_from_GL) > 0]))
        warning ("There are no missing values from GL")

    DT.missing_from_GL <- unique(DT.compare[(times_missing_from_GL > 0) & (gross.oa != 0),  list(times_in_GL=Number_of_times_store_appears_in_GL_for_2013_2014, times_in_OA=Number_of_times_store_appears_in_OA_for_2013_2014, times_missing_from_GL, dates_missing=pasteC(date, C=",")), by=kCols.store])

    ## HAve a look at the example of "Leap Frog"
    DT.compare[gross.oa != 0 & (date > first_date_store_is_in_GL) & storeid==500]
    DT.missing_from_GL


    DT.missing.dates <- as.data.table(DT.compare[(missing_from_GL)] [, table(store_name, date)])

    }, by=storeid]

    stopifnot(DT.compare[, !any(inOA & !inGL)])
    warning("There are ", DT.compare[(inOA & !inGL), lunique(storeid)], " unique stores that have at least one month where they appear in OA but are missing in GL"

    ## show the high gross values

    DT.compare[, diff_in_gross := gross.gl - gross.oa]
    DT.compare[, perc_diff_in_gross := percOf(gross.oa, outOf=gross.gl)]
    DT.compare[, perc_diff_in_gross_by_store := percOf(sum(gross.oa, na.rm=TRUE), outOf=sum(gross.gl, na.rm=TRUE)), by=storeid]
    DT.compare[, mean(perc_diff_in_gross  ), by=storeid]
    DT.compare[, mean(perc_diff_in_gross_by_store  ), by=storeid]
    DT.compare[, mean(perc_diff_in_gross_by_store  ), by=storeid][abs(V1) > 1]