Re: [ddlm-group] Finalizing DDLm

I have no strong feelings either way on the question of rolling or unrolling, though I lean towards Nick and James' view that lists should be enforced for DDLm.  Most files written under DDLm will be prepared under dictionary control anyway.  As far reading legacy files, raading these is surely a matter of implementation, since we know that there will be unrolled loops there.  DDLm can devise its own rules and stick with them.  But I am not experienced in writing this software so maybe I am being too simplistic.

I have just one question.  The loops are currently required to contain a list reference (*_id) that is used to address individual rows in the list.  If a list is 'unrolled', would this item still be required or should a default value be defined in the dictionary (this would strictly be improper)?  Or is it left for the parser do fill in the gap (which might violate some parent-child rules)?  This has some implications for DDLm dictionary construction.

List refernce items are not uniformly supplied in the earlier dictionaries where we (improperly) have relied on the order of the lines, so conversion of the legacy CIFs to DDLm standards during reading into a DDLm application will require list references to be supplied if they are absent.  In some cases the dictionary will be able to construct the correct list reference from the items in the list, but I believe this is not always possible.

David