CSIP80 states that "Each METS file must include ONE structural map element used exactly as described here." and then follows all the requirements for structMap.
The problem with the formulation of CSIP80 is that there is no way to determine if a structMap element is meant to live up to requirements CSIP81-112. There might be two structMap elements which only partially live up to the requirements and how would a validator know which one to validate?
In the testcase https://github.com/DILCISBoard/eark-ip-test-corpus/pull/229/commits/9710696f5581df2183738bd04a7a156f85088b02 I have tried to combine CSIP82 and CSIP80 so that it is the mets/structMap[@LABEL='CSIP'] that determines that this is indeed a structMap-element that it stated to live up to all the requirements.
I suggest a reformulation of CSIP80 to something like:
Structural description of the package
mets/structMap
The <structMap> in the CSIP describes the highest logical structure of the IP.
Each METS file must include ONE structural map <structMap> element used exactly as described in the following requirements regarding the <structMap>
The way to identify the <structMap> element which follows all the structMap requirements is given in CSIP82.
Institutions can add their own additional custom structural maps as separate <structMap> sections.
karinbredenberg
commented
1 year ago
CSIP80 states that "Each METS file must include ONE structural map element used exactly as described here." and then follows all the requirements for structMap.
The problem with the formulation of CSIP80 is that there is no way to determine if a structMap element is meant to live up to requirements CSIP81-112. There might be two structMap elements which only partially live up to the requirements and how would a validator know which one to validate?
In the testcase https://github.com/DILCISBoard/eark-ip-test-corpus/pull/229/commits/9710696f5581df2183738bd04a7a156f85088b02 I have tried to combine CSIP82 and CSIP80 so that it is the mets/structMap[@LABEL='CSIP'] that determines that this is indeed a structMap-element that it stated to live up to all the requirements.
I suggest a reformulation of CSIP80 to something like: