> I have serious doubt that a
native XML database, however well
What does ¡°succeed in commercial space¡± mean to you? XML databases often get their foot in the door by making it easy to store/retrieve the XML that the RDBMS-based applications generate in order to talk to each other. The question is whether the hybrid Object-Relational databases will drive the native databases out of the XML world, not vice-versa. Nobody replaces working systems with new technology and lives to tell the tale ;~)
As for the ¡°failure¡± of OODB ¡ mightn¡¯t this be attributed to the fact that the RDBMS vendors subsumed their unique capabilities into the current ¡°Object-Relational¡± databases, leaving ¡°native¡± OODBs out in the cold? (I¡¯d bet that there are a lot of OODB vendor who would quarrel with the word ¡°failure¡± to describe them, but that¡¯s another matter).
The whole question of how to best store, retrieve, query, and update XML in an efficient, secure, transactional, recoverable, scalable, etc. way is wide-open, and a lot of creativity is being thrown at the problem. From what I can determine from the publicly available material, all the ¡°native¡± XML databases use a rather different underlying approach, and some (like XFinity) even claim to layer ¡°native XML¡± storage on an RDBMS infrastructure. Technological evolution and marketplace competition will sort out the winners and losers ¡ a safe bet is that if Oracle/IBM/Microsoft dominate the XML database market the way they do the RDBMS market today, it will be with hybridized ¡°Object-Relational-XML¡± technology rather than ¡°pure¡± RDBMS technology. For example, it would appear from press reports that Oracle will support some sort of ¡°XML SQL datatype¡± in 9i. That sounds to me like someone in Redwood Shores doesn¡¯t completely buy your vision of ¡°implementing XML data structure on a relational database and map all desired features to RDBMS features.¡±