W3C home > Mailing lists > Public > public-test-infra@w3.org > April to June 2012

RE: Finding specs

From: SULLIVAN, BRYAN L <bs3131@att.com>
Date: Mon, 4 Jun 2012 19:06:55 +0000
To: Robin Berjon <robin@berjon.com>, "<public-test-infra@w3.org>" <public-test-infra@w3.org>
Message-ID: <59A39E87EA9F964A836299497B686C350FEECD8B@WABOTH9MSGUSR8D.ITServices.sbc.com>
Sounds like a great idea. Once you've found the interface or whatever of interest, fanning back out to existing tests for that would make it even simpler to create a new one, using existing related tests as the model.

Bryan Sullivan 

-----Original Message-----
From: Robin Berjon [mailto:robin@berjon.com] 
Sent: Monday, June 04, 2012 9:57 AM
To: <public-test-infra@w3.org>
Subject: Finding specs

Hi all,

last week Dom and I went to a workshop on testing and interoperability for the Web platform in Paris. Part of the discussion was about how to get more community involvement in the production of tests. For instance: you're a web hacker, you find an interop problem between browsers, how do you get to the point where you can submit a test?

We quickly realised that the path from bug to submitted test is way too long. (Some info at http://berjon.com/presentations/20120530-mosquito-testing/mosquito-testing.html). One of the hardest steps along the way (not talking about actually writing a test in the first place) is finding which bloody group to send it to in the first place. That's relatively easy once you've found the spec - but finding the right spec is fraught with peril.

So we had an idea for a small and simple service that would map method names to specs (this is for APIs - the same idea can be generalised to CSS properties and a bunch of other things). The service would do something like:

    * get list of specs from http://www.w3.org/standards/techs/js;
    * extract WebIDL from those;
    * parse WebIDL to extract method names, attributes, etc.;
    * provide a simple search interface mapping those to their containing specs.

There are a few potential issues along the way, but overall it seems doable.

I was wondering if this is something that people here would find useful? Additionally, Dom mentioned there being some existing work along similar lines, so I'd be interested in hearing about that too.

Robin Berjon - http://berjon.com/ - @robinberjon
Received on Monday, 4 June 2012 19:08:04 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:34:08 UTC