Re: HTML Level 2 tests

I agree, I should have stated the question in more detail: what I meant 
is that we should break the big HTML file up into one file per test 
case, and not into smaller files, as you wrote.

/Dimitris
On Tuesday, October 30, 2001, at 06:54 PM, Mary Brady wrote:

>  
>
> 5) Break up the rather large HTML files, which are loaded,
> to exercise the HTML DOM calls, into smaller files, each of
> which will be used for a small set of tests.  It may also be
> possible that we decide in favor of initializing the DOM tree
> by creating these nodes rather than reading them in from
> a file.
>
> [dd] Except if I'm misreading, wouldn't this be a step away from one 
> test - one file that we've used so far
>  
> [mb] Each test reads in an input file.  Currently, all 600 of the HTML
> tests read in the same input file -- which tries to cover many of the
> facets of HTML -- ie, frames, tables, lists, etc.  If a particular
> implementation does not provide all the capabilities of HTML, or
> provides them in a slightly different way, it can be hard to determine
> what's failing and why.  This would be much easier if we broke up
> this huge html file into smaller chunks, and just had tags that were
> pertinent to what we were testing in that file -- or alternately, not 
> load
> a file at all, but rather have an init routine that would construct the
> proper DOM nodes that would correspond to this file -- lets say
> construct a table with 5 rows and 5 columns -- also include table
> header info, table data info, etc that is necessary to exhaustively
> test the DOM HTML Table functionality.  Create another set for
> frames, another for forms, etc.  The actual tests will still be one
> per file.
>  
> --Mary
>