Re: Action items status [DOMTestCase.js]

comments inlined

On Thursday, July 18, 2002, at 09:52  PM, Robert Clary wrote:

>
> There is some use of some DOM methods in DOMTestCase.js however their 
> use over the basic DOM 0 approaches is ok so long as the prerequisite 
> that the user agent implement any required DOM methods/properties such 
> as document.getElementById and HTMLIFrameElement.contentDocument is 
> made clear so that other browser vendors know what will be required of 
> them in order to be able to run the tests in the future.
>
> This includes the use of the non-standard  
> HTMLIFrameElement.contentWindow.
>
> I believe the ability to host the tests on a web site where they will 
> be accessible to all users/developers without the need for them to 
> create the dom ts build environment is important. This includes a user 
> interface for the customization and selection of individual tests. 
> Where do we stand on this ?
>
[dd] We've discussed this in the past and I agree that it is important 
to host the tests/harness somewhere easily accessible. In the past, W3C 
said they could host this, so I think we can suppose they still can 
(otherwise we'll come up with an alternative). The only thing I would 
like to stress is that the hosted version be functionally equivalent to 
the built version, once released, in order to not have differences 
between various DOM TS.

> /bc
>
> Dimitris Dimitriadis wrote:
>> OK, since it seems we have general agreement that the HTML parts of 
>> the TS are stable (except if Rick protests), I'd suggest to point our 
>> energy toward finalizing the framework. Bob, could you please respond 
>> to Curt's request for review and make an agenda for what you want the 
>> rest of us to do to help finish the framework?
>> /Dimitris
>> On Thursday, July 18, 2002, at 08:50  AM, Curt Arnold wrote:
>>>
>>> Dimitris Dimitriadis wrote:
>>>
>>>>
>>>> All,
>>>>
>>>> I'd like to do a survey on how things stand in order to estimate how 
>>>> much more work we need before being able to release the next version 
>>>> of the DOM TS. Most activity of late has been in connection with 
>>>> HTML tests, and not so much the DOM TS framework.
>>>>
>>>> 1. HTML. What is the count of tests for L1/2? Have they been checked 
>>>> for correctness?
>>>
>>>
>>>
>>> I've gotten HTML L1 down to what I think are true implementation 
>>> bugs. Rick was supposed (and probably has) mirrored the changes into 
>>> the HTML L2.
>>>
>>>> 2. Framework. I've informed the DOM WG that we've started working on 
>>>> a new framework and invited the member companies to provide feedback 
>>>> on the new framework. Are there particular issues Bob and Edward 
>>>> would like help with? When do you expect to be able to run the new 
>>>> framework with the old tests?
>>>
>>>
>>>
>>> There is currently a fork in the JSUnit work.  About a month ago, I 
>>> updated test-to-ecmascript.xsl, the support files and provided a 
>>> hacked version of the initial JSUnit 1.3 alpha that would run the 
>>> tests. While I was in the death throes of getting that ready, Bob 
>>> issued a revised JSUnit 1.3.  I had asked Bob to review the generated 
>>> tests and my hacked JSUnit and suggest a path forward, but didn't get 
>>> any feedback.  I like the way the tests are currently rendered and in 
>>> general, like the modifications that I made to JSUnit.  However, I 
>>> did the JSUnit mods quick and dirty and may have broken something, so 
>>> I'd suggest incrementally working in the changes and making sure that 
>>> existing non TS tests still run.
>>>
>>>>
>>>>
>>>> There are a few more issues that I've taken up with the DOM WG, in 
>>>> particular:
>>>>
>>>> 1. Tests for particular modules. We've discussed having tests 
>>>> written for the modules by those responsible for producing them (the 
>>>> member companies whose representatives prdocue the specifications).
>>>> 2. DOM TS Group involvement: I've raised the issue of not having a 
>>>> very balanced division of labour (as indicated earlier in a series 
>>>> of postings to this list) which in turn means that we cannot ensure 
>>>> that all implementations will run smoothly with the new framework.
>>>>
>>>> Any other issues?
>>>>
>>>> /Dimitris
>>>>
>>>>
>>>
>>>
>>>
>>>
>
>
> -- Bob Clary, [email protected]
> Technology Evangelist, Netscape Communications
> http://developer.netscape.com/evangelism/
>