<burn> scribenick: tibbalst

Describe plan for the call

Yes. Didn't realize you wanted in window.

WIll start.

Test Suite Issues and Discussion

<burn> Test suite: https://github.com/w3c/vc-test-suite/issues

isaac: (invited guest for today's call) - first attendance. Bloom: Credit and Identity Protocols. Updated recently with spec. Want to get feedback on their implementation and make sure it’s interoperable. - Have a detailed spec. Can send email to public mailing list. - Can listen in on call.

yancy: Yancey Dialed in by phone.

dmitriz: Have a couple of documentation issues from brent.

<dmitriz> from brent: https://github.com/w3c/vc-test-suite/pull/27 and https://github.com/w3c/vc-test-suite/pull/25

<dmitriz> and corresponding issues: https://github.com/w3c/vc-test-suite/issues/26 and https://github.com/w3c/vc-test-suite/issues/24

davidwaite: New attendee. Just listening in.

dlongley: will make comments on PR to make changes.

jonathan_holt: Trying to build specification to validate that all the attributes are there and all qualified values. Where would I find repository of appropriate values for field?

dmitriz: short answer. Schema is supposed to be open ended.

<dmitriz> https://github.com/w3c/vc-test-suite/issues/23

dmitriz: issue 23 Points out by andrew jones. Utils file needs a copyright on top.
... will add copyright. Assign me.

<dmitriz> https://github.com/w3c/vc-test-suite/issues/22

dmitriz: issue 22 by brent.
... changed the spec in March from ISO to RFC.
... had to figure out how to convert back. volunteer to be assigned

<dmitriz> https://github.com/w3c/vc-test-suite/issues/21

dmitriz: question on how to increase time outs by brent.
... something missing in config file. brent wants to override the timeout.
... I will add timeout to readme.

<dmitriz> https://github.com/w3c/vc-test-suite/issues/19 and https://github.com/w3c/vc-test-suite/issues/18

dmitriz: 2 issues on same thing by yancy.
... W3 id serving the context is not returning the JSON context. Throwing an error. Need someone to update content type to application/json-ld

<dmitriz> just to clarify: W3 id is serving the context, but is returning the wrong content-type header for it.

yancy: Was able to get around the issue.
... used local file.

<dmitriz> also I believe issue https://github.com/w3c/vc-test-suite/issues/9 by yancy is about the same thing (plus a CORS header)

burn: need update on content type.

<dmitriz> propose we assign issues 9, 18 and 19 to kaz.

kaz: will update.

<dmitriz> next up: https://github.com/w3c/vc-test-suite/issues/14

<burn> I just assigned kaz to those

<dmitriz> propose close: https://github.com/w3c/vc-test-suite/issues/2

dmitriz: last issue. old one from when test suite was first being planned. Propose to close it.
... Chris webber thinking outloud on the issue.

burn: recommend add comment in before closing.
... please close after adding comment.

jonathan_holt: Trying to understand more on the issue.

dmitriz: Trying to structure verification which is not implemented in test suite.
... last of issues.

Implementation topics discussion

oliver: no implementation topic but on test suite. Wanted to ask that implementers will create wrappers for library?

dmitriz: topic brought up in Barcelona. Answer was yes.

dlongley: Build driver on top of test suite. Pass any information to library. Additional Layer for most part. Some driver will receive information.

oliver: Please add diagram in test suite to explain this?

<ken> +1 to adding a diagram

<brent> +1

dlongley: Need volunteer to make diagram.

oliver: Started one and may be able to provide. Oliver volunteered.

yancy: Wanted to ask /clarify on bad cardinality. If using Json-LD parser, what are the algorithm that figures out if only the base context when it should fail. It is missing context. It could be done in different ways besides using context. Looking for clarification.

<Zakim> dmitriz, you wanted to answer yancy

dmitriz: if using json-ld processor Will just throw an error about missing context.

<Zakim> dlongley, you wanted to answer

yancy: Want to make sure we are doing the same way.

dlongley: Updating PR that context field needs to be checked by all implementations.
... Additional constraint that it must be present and have certain values in it.
... Don't think the spec must have two context.

yancy: it does.
... Spec says to check explicitly.
... Can look at PR and review data model to understand better.

ken: how to get to optional tests to invoke?

dmitriz: short answer is the optional tests that are being skipped in the queue are not implemented. Expect sovrin community to implement. Not sure where to start?
... Will need to implement them if you want to invoke
... There are a couple of tests that are skipped regarding term dictionary. Question to group if that should be a test and if testablel

oliver: Question on short tests themselves. [could not get question]

burn: there are no normative statements on use of 'vp' in JWT. What tests can there actually be. The only test can be existance and that it has contents.

dmitriz: Dave can go into more detail. Basic and advanced tests do cryptography checking. Should it check signature?

<Zakim> dlongley, you wanted to respond to oliver

oliver: Fine with also writing the JWS test(s).

dlongley: Even if there are things that are non-normative would be great to have tests to check

oliver: If the JSON-LD contains issuer or issuance date property, the JWS test would check to see if they are transformed correctly.

dlongley: having an optional test is good.

brent: If fully conformant VC, would that credential fail test suite if fields do not have the data. Should test suite fail it?

dmitriz: Please raise if there is an issue if optional fields are blank and it's raising an error. Maybe a warning.
... Up for discussion if it should be warning.

brent: Hope is that it shouldn't test suite should not fail.

<Zakim> ken, you wanted to ask about reporting

brent: Will raise issue if actually a problem.

ken: Question: unimplemented feature like terms of use. If we have items that we are not using do we just let test fail?

dmitriz: good question.

burn: ALways help to report which ones you did not implement. Big difference between failing and not implementing test.

dmitriz: Helpful to us to split the test invocation commands to run the tests and the optional tests, terms of use tests, etc. That way there may be less confusion on which ones are passing.

<brent> +100

<Zakim> jonathan_holt, you wanted to discuss recursive approach

<ken> +1 Selective test running will also speed up the testing for me.

jonathan_holt: Can break down into chunks. Can implement test if needed.

dmitriz: Agree with JSON schema Valid approach. Only reason to split out because of the way test suite is structure. One of the tests... Send in terms of use and see what it does. If library doesn't handle it will mark it as fail.
... Need to separate out tests.

jonathan_holt: Will be breaking out tests. Will give some feedback.

burn: support for selective test running.

oliver: Optional vs. Mandatory. Will be difficult to identify what should be in each. (One proof must be present. but proof property is optional). So what is the test?

dmitriz: Thought change that proof property was required.

burn: for Verifiable credential is required to make it valid.

dmitriz: Answer question?

oliver: Will need to check specification
... Proof property is not specified as required.
... Is there a test that proofs that it's there?

dlongley: Can you clarify?

oliver: How do we show compliance?

dlongley: Compliance is based on report if passing. [what was solution?]

dmitriz: Tests are always checking proof property. Need way to specify that they have way of external proof.

<ken> +1

<brent> +1

<oliver> +1

burn: valuable?

<yancy> +1

<dlongley> +1

<Zakim> ken, you wanted to ask about preliminary vs. final reporting.

burn: Will allocate more time for these types of conversations?

ken: Report as soon as you have it?

dmitriz: Report to PR in implementation directory.
... if updates.

<Zakim> burn, you wanted to remind about testing spec, not implementations

burn: is the goal of this to test the specification, not implementation? No one wants to submit that there implementation is inferior. Marketing departments should not use the results to compare. In the implementation report to test specification. If you want to do a conformance test for implementation it would require WAY more tests. A failure should not be viewed as a failure of the implementation.
... question?

oliver: Mobile drivers license standard is about to be finalized. Have opportunity to provide how VC and they could align/cooperate. What needs to be changed in each standard to work together? Should use opportunity.

DavidC: need to submit comments for meeting in July (in Portugal). Have made suggestions. There is an issue that MDL specifies protocols. There needs to be a protocol. To be useful, we need to discuss protocols.

burn: will not have group decision by Thursday. Chairs can discuss going forward.
... Need two hour timeframe due to issues. Talk next week.

[adjourned]