I already run queries against this dataset on weekly basis . If this is bug I prefer it to be fixed over engineering resource intensive workaround it.