On the one hand, there are various issues with the Geoserver’s DCAT-AP export: e.g., they use non-existing hydra IRIs for the pagination, they have some prefixes undefined (leading to broken IRIs), the pages take very long to respond - they even sometimes time out, and the distributions are blank nodes instead of given an IRI. Can we get in contact with Geoserver to solve that? Is there a reference to their code that generates this to maybe do a pull request there?
On the one hand, there are various issues with the Geoserver’s DCAT-AP export: e.g., they use non-existing hydra IRIs for the pagination, they have some prefixes undefined (leading to broken IRIs), the pages take very long to respond - they even sometimes time out, and the distributions are blank nodes instead of given an IRI. Can we get in contact with Geoserver to solve that? Is there a reference to their code that generates this to maybe do a pull request there?
On the other hand, the scripts provided in this repository at https://github.com/Informatievlaanderen/vodap/blob/master/.circleci/config.yml and https://github.com/Informatievlaanderen/vodap/blob/master/scripts/download.sh also contain an important mistake: the parser re-starts numbering blank nodes on every page, so the blank node numbering across pages will conflict.
I’ve writen a small nodejs script that does number blank nodes correctly over here: https://github.com/pietercolpaert/DCAT-AP-Dumps-To-Feeds/blob/main/bin/helperFlanders.ts - feel free to reuse it here.
Needs input from @bertvannuffelen