Handling of streaming sources - Was: Why is "minimum one file in a geospatial format" only SHOULD?

[anitagraser]

p.21 Why is this requirement not must?

[GregorZavrsnik]

Since all general and content-type specifications for preservation need to allow for maximum adoptability, it was the decision of the DILCIS Board, that general specifications should be as open as possible, and therefore avoid use of MUST criteria for validation as much as possible. Especially, since we didn't agree or prescribe what the final list of geospatial formats are and can therefore not be sure if we missed some.
Generally, it is up to the end-user to upgrade this criterion to a stricter level in their local implementation if they deem it fit for their purpose.
Although for most organizations this isn't logical there could be examples, where this is the case
For example, if an organization decides, that they want to store geospatial data in an RDBMS type format (SIARD, Oracle dumps, etc.), they can still classify that archival package, containing a database as a geospatial record, if that is how they use it. Although the reusability of such a package might be reduced across a wider audience, the aim is to support as many organizations as possible.

[GregorZavrsnik]

After multiple discussions with different Geospatial organizations, it was pointed out, that Geospatial information products are often created using only online sources from other organizations. This therefore poses a problem, how an organization should transfer the creation of their records to the archive if the sources are mostly streamed from services.
So this issue required further investigation.