RE: In a quick glance, all of the whitespace test file are invalid tests that don't prove anything

Melissa Newman wrote: 
> There is a major problem with the whitespace test files (all of them).  The
> problem is that you can't tell from the results (or even the original data) what
> the inputs were and what the outputs should be.  For all whitespace testing,
> we should be using the &###; format instead of trying to insert a space, tab
> or whatever. 

There is no need to use &###; though it might be more helpful to understand when looking at the code the actual character itself is just as valid.

> Plus, the line right below it, should indicate with a color block
> what the width of the final whitespace is supposed to be.
> 
> If I have a test that says :
> X     X
> 
> "The results are two Xs", how is an end tester going to see the difference
> between:
> 
> X    X
> 
> And
> 
> X X
> 

If you are seeing the letter 'X' then you need to install the ahem font. These cases rely heavily on that font to properly test white spaces. Also many of the white space cases may require a special font in the future just for testing white space. Make sure you have properly installed any necessary fonts called out in an prerequisites of the white space tests (if there are any at the moment).

> Plus, how do I know if the character there is a single white space, a double
> white space, a zero white space, or a control character?

One final question. Could you provide a couple of specific links to files you are concerned about? I see nothing incorrect with the white space testing methods at the moment but maybe we are looking at different things.

--
Thanks,
Arron Eicholz

Received on Tuesday, 2 March 2010 15:09:33 UTC