DILCISBoard / SIARD

SIARD (Software Independent Archiving of Relational Databases) - an open file format for the long-term archiving of relational databases
10 stars 0 forks source link

Optional/mandatory element: Databaseproduct #23

Open PhillipTommerholt opened 5 years ago

PhillipTommerholt commented 5 years ago

Review comment:

5.1 Database level metadata Why is “DatabaseProduct” not mandatory? This information is essential - especially when it comes to interpretation of datatypes. Could “databaseProduct” as a minimum be mandatory when “typeOriginal” is included in metadata?

PhillipTommerholt commented 5 years ago

It is currently not known to the SIARD Working Group why this element is optional and not mandatory. There might be situations where table data can be retrieved, where no information about the original database product is known?

The SIARD file format specification is generally very open with many optional elements. See some of the vital optional elements:

For preservation purposes it is often wise to make sure that the examples mentioned above are mandatory (to the extend that they can be retrieved) in order to make sure that the interpretation of data will also be possible in the future and that the connections in the data today will also be documented. The SIARD Working Group encourages users and institutions to have their own stricter and narrower SIARD-profiles that lie within the possible valid outcomes of SIARD file format specification. The SIARD Working Group will focus on possible changes to the specification where users can not preserve their significant properties due to the limits of the specification, rather than limiting the possible outcomes within the specification since the SQL standard in itself is very wide. Many of the questions regarding stricter profiles, extra metadata and the SIARD-file within an Information Package will be dealt with in the work for an E-ARK Content Information Type Specification for Relational Databases.