The Forum for Discussion about The Third Manifesto and Related Matters

You need to log in to create posts and topics.


Some of you will no doubt have seen this, but if not, it might be of interest:

I'm the forum administrator and lead developer of Rel. Email me at with the Subject 'TTM Forum'. Download Rel from

Thanks Dave, like everybody (it seems) I can barely contain my apathy.

blah blah SQL blah SQL blah Don Chamberlin blah SQL ... I'm sure it'll do very well.

One of the world's largest aggregators of commercial data wants your enterprise to hand over all its data in any old format. For free -- so it must be that your data is the product. What could possibly go right!

I did learn a new term 'data lake'. But it seems all the best wisecracks have already drained it. Global warming, I suppose.

In other news, the patent on Bachman's participant data model, the last gasp of Codasyl, expired in the U.S. a few days ago.  So I read it over, and after removing the patentese and Bachmanese, I think what it is is a graph database (that is, one in which some attributes are typed as containing a synthetic primary key).  The only difference is that conventional graph databases have one-way links between nodes, whereas the participant model makes all links (called "participants") two-way.  That obviously can be layered over the one-way links, but also allows more convenient navigation.

I also read up on Bachman's earlier role data model, which is somewhat more interesting to me (but I like history and alien ways of thinking).  Rather than a record type being defined directly by its attributes, the attributes belong to role types, which are then aggregated in a flat way into record types.  However, the Codasyl owner-members relationship (miscalled a "set"; it is ordered and allows duplicates) between record types is replaced by one between role types.  This allows considerably more flexibility: in particular, it subsumes non-Codasyl extensions whereby a "set" can contain members of various record types.