Re: XHTML Test Suite Minimal document

I'm sorry to have not been very active in this project, but a number of
pressing matters came up.  I'll try to be more timely in future.  

If you would like to have a look, I have some material at
    http://www.math.ucla.edu/~jimc/html40-test/index.html
I went through the HTML-4.0 spec and tried to put together one of
everything.  Mostly I bypassed CSS style effects in favor of deprecated
tags and attributes, though a few CSS effects are illustrated.  

As the intended audience was not the worldwide community, my texts lack
some items (like section references for everything) which in retrospect I
probably should have put in (and probably will put in).  Also, I suspect
that I have identified some tags as inline when the standard actually
intends them to be block level.  And the material tests HTML-4.0 and has
not been updated for XHTML.

Shane's tool sounds good, in which the test document is actually a form and
it is backed by a CGI that accumulates user judgements on each test.

However I see one problem with that method: if the test suite is hosted on
W3C's server the CGI deposits the results on W3C's server, to which the
user has no access.  You would have to download the test suite (with its
CGI) and install it on your own server, which is completely appropriate for
a developer such as Netscape, but which couldn't be managed by students or
third parties evaluating browsers.

I wonder if this would be possible -- I'm no Java expert so I suspect it
would work but don't know.  The toplevel page of the test suite takes you
into a Java applet (or better, ECMAscript/Javascript, because paranoid
people avoid Java but may be convinced to tolerate Javascript for an
important purpose).  In a text field on the toplevel page (or put up by the
applet) the user provides a mail address.  The applet persists throughout
testing (assuming the browser doesn't crash, a definite possibility in a
test suite) and interprets "submit" button presses on each page, mailing
the form content to the user, one message per page.  It's nasty, but it
gets the results off the web server.

As an alternative, conceivably the applet could be given the name of a file
on the browser machine, but no sane person would voluntarily allow Java to
write, much less read, an arbitrary file.

James F. Carter        Voice 310 825 2897	FAX 310 206 6673
UCLA-Mathnet;  6115 MSA; 405 Hilgard Ave.; Los Angeles, CA, USA  90095-1555
Internet: jimc@math.ucla.edu (finger for PGP key)
UUCP:...!{ucsd,ames,ncar,gatech,purdue,rutgers,decvax,uunet}!math.ucla.edu!jimc

On Thu, 14 Dec 2000, Shane P. McCarron wrote:

> That is how our current framework works (the ApTest one). For each test,
> you select a result and provide notes.  The notes are accumulated in a
> test-session specific bucket so that they are 1) private to the user,
> and 2) you can see how the results evolve.

Received on Thursday, 14 December 2000 13:41:08 UTC