In addition to the global metadata and specific variable metadata the header
must also specify the delimiter to be used to separate data entries. This
delimiter is the same for all variables (but may differ from that used in the
header for attributes), and is placed with the global metadata. It has
syntax
Data_delimiter = x
where x is any printable character. If no entry is specified
Data_delimiter =
a space character is used. This is also the default character if no Data_delimiter line is present.
In delimited files the parameter ``Ignore_on_read'' may be specified within any variable meta data block. This parameter takes the value ``yes'' (or ``YES'') to indicate that this variable is not wanted. All other values are ignored. If this parameter is set then no other information need be specified, and the variable name may be a dummy entry. If the ``Sizes'' parameter is set this will be used to determine how many columns to ignore in the data file. It provides an easy way to skip columns of data in files generated by packages other than QSAS. There is no requirement that columns so skipped should be of the same data type, merely that each column has the required delimiter.