Redundant availabilities

[margaretha]

If a registered user requests a search, then changes a connection to VPN afterwards and resends the search, the availability rewrite will result in

This could be improved with a rewrite revert mechanism as described in the Kalamar issue KorAP/Kalamar#65.

[margaretha]

I have made some thoughts about the issue. I think the current approach with redundant availability rules is a safer and transparent way of rewrite.

Virtual corpus (Collection) Rewrites that have been done in a request is not considered in the next request. This is correct, because the next request includes the availability rules in a corpus query.

In fact, user sends the next request with the corpus query containing availability rules, even though they came originally from rewrite mechanism.

Merging new the availability rules is less safe because it may disregard a rule

for instance:

A request with login and corpus query

availability eq QAO-NC

this will be rewritten into

availability eq /CC.*/
or availability eq /ACA.*/
or availability eq /QAO-NC/

That would retrieve more documents than required.

It would be easier and clearer if the query and corpus query are simply reset after login/logout. VPN change is more complicated to detect...

[Akron]

I don't think I understand. So - you would argue that KorAP/Kalamar#65 is not prefered?

[margaretha]

The suggestion in KorAP/Kalamar#65 to revert the rules is good! I was trying to describe a possible problem, if this case were to be handled in Kustvakt. This is much easier to handle in Kalamar. So I think we can close this issue.