the feature of ifc5+

[j209]

Hi
I think the next generation of ifc should be based on the database and not the file. like the imodel concept that was developed by Bentley and is based on the sqlite database. It is used as the core of data storage and changes in a physical assets and digital twins, and it can be a good idea for those involved in the development of ifc 5+.

[aothms]

I understand where you're coming from. I've also experimented with several kinds of mappings to sqlite. But I would phrase it differently though. We should move towards a data model that has a straightforward representation in a relational database. So that we can leverage these kind of features. But the main predominant format would probably still be something file-based because it's the most neutral kind of exchange format. The ECS prototype here, because it's a sequence of much more homogeneous atomic data fragments is already much more easily represented in a DB. This prototype is outdated though because it does not have any means for "composition behaviour", the functionality that inspired us to look into USD.

[j209]

using relational databases as the basis for storing ifc information may not be best solution due to the complex and multi-model nature of the ifc files. My suggestion is multi-model databases like Surrealdb.
see This.