First thoughts on testing

I think it will be extremely valuable to readers of the guidelines to
know how well constructs such as bidi, vertical text, lang selectors,
etc. are supported by browsers.  Not only will it be practical
information for them, but it will help us to include information about
future directions or sparsely implemented features that may still be of
use if you are dealing with a known browser-specific population (such as
people in large corporations do for internal information). To assemble
that information, however, will require us to do testing.

Some points that to my mind are worth discussing include:

-	we probably can't test all browser versions on all platforms,
and in many cases that wouldn't be worthwhile anyway, since people don't
use some of the older ones - so which ones should we be concerned about?

-	reporting on old quirky browsers may take the readers gaze away
from the standards route - I don't 
think we should put too much focus on workarounds for non-standards
compliant browsers

-	new browsers have different modes: quirks, almost standards, and
standards - we should probably take that into account, although we
should probably also try to convince people to go with standards mode
wherever possible

-	people can serve documents as HTML or XHTML; transitional,
strict, etc; text/html, application/xhtml+xml; etc.  How far should that
concern us?




My first thoughts on how to move forward include:

I think there is significant value to reflecting summary information in
table form in our outline documents (see
http://www.w3.org/International/geo/html-tech/outline/html-authoring-out
line.html ), but to make it work we need to choose a manageable yet
useful set of baseline browser configurations and versions.  We should
then test against these baseline configurations and key subsequent
releases.

Remember that we'll also have the ability to throw in descriptive
information in the narrative about problems with specific browser
versions, and this could go well beyond our baseline set.

I'm planning to use a simple mechanism in the table based summaries to
draw attention to bugs and special features that are then described in
the narrative.




I would suggest that we use the following as baseline configurations:
Opera 7 (standards mode)
IE Mac 5 (standards mode)
IE Win 6 (standards mode)
Navigator 6 (standards mode)
Mozilla 1.0 (standards mode)
Safari v?

I suggest that we additionally test against:
Navigator 7 (almost standards mode)
Mozilla 1.1, 1.4? (almost standards mode)

We could also consider testing against Navigator 4.? and IE 5 or 5.5,
and IE6 with xml declaration, though I see these as a lower priority,
and the tests as fodder for the descriptive comments rather than the
summary tables.


Thoughts?
RI


============
Richard Ishida
W3C

tel: +44 1753 480 292
http://www.w3.org/People/Ishida/ 
http://www.w3.org/International/ 
http://www.w3.org/International/geo/ 

See the W3C Internationalization FAQ page
http://www.w3.org/International/questions.html

Received on Wednesday, 6 August 2003 13:07:57 UTC