- From: Marcos Caceres <w3c@marcosc.com>
- Date: Mon, 14 Jan 2013 16:48:59 +0000
- To: Clint Hill <clint.hill@gmail.com>
- Cc: Brian Kardell <bkardell@gmail.com>, "public-nextweb@w3.org" <public-nextweb@w3.org>
On Monday, January 14, 2013 at 4:40 PM, Clint Hill wrote: > Marcos: I'm glad you showed this as I just committed a Boolean test > fixture. I also used AMD. But we have different approaches (I'm using > require and you use define). I put my test in > "web-idl-type-boolean-tests.js" in the /test directory. > > I think my pattern is a little lighter than yours and might make test > writing a bit easier? I actually started there, but after a few hours switched to using define as I started hitting problems. The problem was that you then needed another file to load in all the interfaces. Like: require(["type/Boolean", "type/Byte", and so on]) Plus: require(["test-file-1.js", "test-file2.js"]) Using define, overcomes this problem and makes everything self contained. I also found other people recommending this approach. See, for example: http://www.jonnyreeves.co.uk/2012/qunit-and-requirejs/ > I'm happy contribute to the test writing as maybe that's a way to get > blind taste tests since I didn't write the implementations? That would be great. But please make sure you copy/paste from the WebIDL spec as I've done. It helps make sure that we have a 1-to-1 test suite and implementation. > Maybe that > will create a cleaner validation process of the implementations this way. > > Thoughts? -- Marcos Caceres
Received on Monday, 14 January 2013 16:49:35 UTC