# This information should be included somewhere in the documentation
# for the Miro libraries. Since the error-reporting functions reside
# in the parse library, perhaps it would be appropriate there.

Two quirks in the way errors are reported should be explained.  First,
if an entry has more than one semantic error associated with it, all
those errors may not be reported. If the input is fixed to remove the
error reported for that entry, other errors in that entry may be
reported on subsequent parsings.  Second, since the input is read but
not processed in sequential order, the errors are not necessarily
reported in increasing order by line number.  Instead, all errors
associated with BOX entries are reported first, followed by those
associated with ARROW entries, followed finally by those associated
with INSIDE entries. To see the errors in increasing order by line
number, pipe the standard error output through sort(1) by appending
"|& sort -n" to your command line.
