pyshacl is failing because it hitting the limit of the number of tests it will run. This happens not just because we have a lot of tests but because there can be many ways of traversing the graph to reach a point where a test is run, so each test may be run many times.
It's not enough to have smaller test data files because this isn't an option for data providers: they need to provide all the relevant data.
A better solution is to have several SHACL files, one for each "subdomain" of the data, e.g. one for Organization & related classes; one for Person details; one for Course-related data. This more modular approach may well have advantages if other PESC standards switch to linked data as they can reuse the relevant modules.
To get several SHACL files we need several coresponding TAPs.
To batch run the tests on the SHACL we need subdirectories of the TestFiles for each of the "subdomains" that has a SHACL.
Start by factoring out:
[x] Contact data (ContactShape, PostalAddressShape, PhoneShape)
[x] Organization data (OrganizationShape, OrganizationIdentifierShape)
[x] Person data (StudentShape, NameShape, ResidencyShape)
pyshacl is failing because it hitting the limit of the number of tests it will run. This happens not just because we have a lot of tests but because there can be many ways of traversing the graph to reach a point where a test is run, so each test may be run many times.
It's not enough to have smaller test data files because this isn't an option for data providers: they need to provide all the relevant data.
A better solution is to have several SHACL files, one for each "subdomain" of the data, e.g. one for Organization & related classes; one for Person details; one for Course-related data. This more modular approach may well have advantages if other PESC standards switch to linked data as they can reuse the relevant modules.
To get several SHACL files we need several coresponding TAPs.
To batch run the tests on the SHACL we need subdirectories of the TestFiles for each of the "subdomains" that has a SHACL.
Start by factoring out:
Others can be factored out as we write tests for them. Probably will be