- From: Michael(tm) Smith <mike@w3.org>
- Date: Tue, 1 Dec 2009 01:35:17 +0900
- To: Jonathan Griffin <jgriffin@mozilla.com>
- Cc: Kris Krueger <krisk@microsoft.com>, "'public-html-testsuite@w3.org'" <public-html-testsuite@w3.org>
Jonathan Griffin <jgriffin@mozilla.com>, 2009-11-30 11:58 -0800: > It's hard for me to wrap my head around what kinds of metadata we'll need > and how it should be stored without seeing the larger picture. We know > there are existing tests for many parts of HTML5 in various places; I think > it would be helpful to collect these on a wiki somewhere for review: > > - how suitable are they for inclusion into an official HTML5 test suite? > - how much work will they take to cleanup and extend? > - if they're not automated, will we automate them, and how? > - will we try to tie them all together with a common test runner (e.g., > http://omocha.w3.org/wiki/)? > - do we want to manage storage of test results, e.g., in a SQL or CouchDB > server? > > Having the answers to these questions might take care of some of these > metadata questions, or at least provide some framework for answering them. To give my own feedback on some of those questions: I think we should make it a goal to automate all tests. Or at least make it a goal to separate tests into sets that are automated and sets that aren't, and to make it goal to keep the sets of non-automated tests to an absolute minimum. I think we should try to tie tests together into a common test runner or common group of test runners. I can imagine that we might want to have, say, Sylvain Pasche's browsertests cross-browser test-runner be among them, as well as the cross-browser test-runner system that the Microsoft testing team has put together. The thing that would unite that is that they would implement a common client API for submitting test results back to our central test-results repository. On that note, I do also think that yes we do want to manage storage of test results in a central database of some kind. And in order to be able to do that, the very minimum piece of metadata we need is a unique identifier for each test case. --Mike -- Michael(tm) Smith http://people.w3.org/mike/
Received on Monday, 30 November 2009 22:50:42 UTC