Re[3]: [cgmo-webcgm] Sample test page

[Benoit Bezaire <benoit@itedo.com>]

>>[...]
>>DL>   >   - We need to agree on the size (in pixels) of the CGM illustration,
>>DL>   >   how big should we make it?
>>
>>DL>     512x512 seems reasonalble to me, e.g. 1/2 of the typical
>>DL>     screen height.
>>I think the test pages should fit into a 1024x768 display. So
>>something a bit smaller than 1/2 sounds good (we have to take into
>>account browser borders and separators). How about 480x360? (to
>>maintain the aspect ratio).

LH> Just to be clear ... we're talking about 'width' and 'height' attributes on
LH> <object>, right?
Yes.

LH> I don't recall any discussion of the <param> child element, and its
LH> attributes (for aligning, fitting, etc).
I don't think <param> is required for us to develop a test suite.

LH> Has anyone implemented these yet?
I don't know.

LH> Should we avoid them in the DOM suite
Yes, in my opinion.

LH> (but ... they ought to have a test somewhere, right?).
That's not a WebCGM 2.0 test, that's a version 1. The 1.0 test suite
should have been testing that.

>>DL>   >   - Where should the reference image go? On the right side?
>>
>>DL>     Sounds ok.
>>Ok.
>>
>>DL>   >   - What happens when a reference image is not required, like in this
>>DL>   >   case?
>>
>>DL>     Leave it to test writer.
>>We should probably agree on something, all pages should have the same
>>structure.

LH> I strongly agree with that.
Ok.

>>DL>   >   - What do people think?
>>
>>DL>     I would like to see a test name at the top of each test page.
>>Good idea.

LH> Are we able to list the potential components of a test page?

LH> -- display of the CGM (result of processing <object>)
Ok.

LH> -- title
A title is good, Ok.

LH> -- buttons to execute tests (always?)
I don't think it needs to be _always_, but we could probably make it
_always_.

LH> -- reference image (only sometimes?)
Not always needed (in my opinion), but we can have one anyway, that is
the same as the source.

LH> -- text results (only sometimes?)
Only sometimes (I guess), the mouse interativity tests will be
difficult to expressed in text.

LH> -- text instructions / operator script (sometimes?)
We can make both of them _always_. In some cases it be as simple as
'look for pass or fail in the table above'.

LH> I guess the trick is that some bits are present always, some are present
LH> sometimes, etc.  Are reference image and text results mutually exclusive,
LH> or might some tests have both?

LH> Anyway, a prescription for uniform layout of the potential test case bits
LH> will be good for everyone (test authors, users, etc).
Yes, I agree.

LH> -Lofton.

-- 
 Benoit   mailto:benoit@itedo.com