- From: Markus Lanthaler <markus.lanthaler@gmx.net>
- Date: Sun, 25 Aug 2013 13:02:40 +0200
- To: <public-linked-json@w3.org>
On Saturday, August 24, 2013 11:47 PM, Gregg Kellogg wrote: > I updated the test-suite instructions with this, and modified the > vocabulary. The vocabulary's maintained in > http://json-ld.org/test-suite/vocab.ttl. There is a small ruby > script to generate vocab.jsonld and vocab.html (in HTML+RDFa) from > this. You can see the HTML+RDFa version at > http://json-ld.org/test-suite/vocab. Between the two documents, developers > should have everything they need to know to create a test runner to run > the test suite. Great work Gregg. I have a couple of comments but I think it's easier to discuss those things directly in the commits. Thus I added them there [1]. > As the runner should not have any a-priori knowledge of the test > manifests, we can probably simplify the test organization. I'd like > to eliminate error-expand-manifest.jsonld and fold that test into the > expand-manifest.jsonld. It also uses an @type of jld:ApiErrorTest, but > this is redundant with jld:NegativeEvaluationTest, which already expects > an error result. While I fully agree that the runner should not have any a-priori knowledge of the test manifests, I think keeping them separate simplifies the creation of test runners a whole lot. You don't need to generalize the test methods in that case but can keep them very short, a couple of lines in most cases (here's how I do it [2]). It also makes it much simpler to run just a subset of the test suite. I would thus prefer to keep them separate. > Next task is to identify tests; I suggest that we create manifest entries > for any MUST clause not already having a test (if we were really ambitions, > we'd annotate each test with what normative requirement it is testing). Yeah. I would also suggest to go through the algorithms and write a test for every error that is thrown. > If we take a divide and conquer method, we should be able to get through > these in the next week or so; certainly, prior to entering CR. I'm quite busy at the moment but I'll certainly find the time to write some. Just tell me what subset of the spec I should work on. [1] https://github.com/json-ld/json-ld.org/commit/73f2c63f0c37f04f46bda17da3f6ac fe321b282a [2] https://github.com/lanthaler/JsonLD/blob/master/Test/JsonLDTestSuiteTest.php #L69 -- Markus Lanthaler @markuslanthaler
Received on Sunday, 25 August 2013 11:03:16 UTC