Https urls currently not allowed in conformance checks

[fjugipe]

The spec says:

Which means that both http and https URLs are valid, as long as the API adheres to the respective protocol's standards. Maybe the tests should check both with for example:

conformance_urls = [
    "http://www.opengis.net/spec/ogcapi-common-1/1.0/conf/core",
    "http://www.opengis.net/spec/ogcapi-common-2/1.0/conf/collections",
    "http://www.opengis.net/spec/ogcapi-edr-1/1.1/conf/core",  # Assume this will be changed to edr_version
]
openapi_conformance_urls = [
    "http://www.opengis.net/spec/ogcapi-edr-1/1.1/req/oas30",  # Assume this will be changed to oas31
    "http://www.opengis.net/spec/ogcapi-edr-1/1.2/req/oas31",
]
...
def requirementA2_2_A5(jsondata: dict) -> tuple[bool, str]:
    OGC API - Environmental Data Retrieval Standard
    Version: 1.2
    Requirement Annex A2.2 A5

    Check if the conformance page contains the required EDR classes.
    jsondata should be the "conformsTo"-part of the conformance page.
    spec_url = f"{edr_root_url}#req_core_conformance"

    if "conformsTo" not in jsondata:
        return False, (
            f"Conformance page does not contain a "
            f"conformsTo attribute. See <{spec_url}> for more info."
        )

    for url in conformance_urls:
        http_url = url.replace("https://", "http://")
        https_url = url.replace("http://", "https://")
        if http_url not in jsondata["conformsTo"] and https_url not in jsondata["conformsTo"]:
            return False, (
                f"Conformance page does not contain "
                f"the core edr class {url}. See <{spec_url}> for more info."
            )

    return True, "Conformance page contains the required EDR classes."

[fjugipe]

The same applies for requirementA11_1

[havardf]

@fjugipe Thank you for the feedback. We need to look into this. The actual conformance link to EDR is not quite correct either, since we target EDR v1.2, which is still in DRAFT, and do not have a canonical conformance class url yet, as far as I can tell.

How much of a blocker is this for you at the moment?

[fjugipe]

Not a blocker! Just had me stalled for a while when using the tool against eumetnet edr until I noticed what was the issue.

[geira]

I think this is mixing up two different parts of the standard into something that was not the intention. Req 1B (§ 8.1) only says something about what protocols an API should support, nothing about links in conformance. Remember that those are not URLs but URIs (actually IRIs), meaning they are unique string IDs which may or may not happen to resolve to a web page as a side effect (indeed the OGC http conformance links end up going to https://docs.ogc.org/). This means "http:..." and "https:..." URIs are NOT identical, and there should be no implicit conversion between protocols.

Each spec/conformance class should have ONE (and only one) canonical unique identifying URI, optimally implemented with a redirect from http to https on the web server (something which is a legal requirement for public web sites in Norway, possibly also in the EU).