FW: [Webtest] Forms and image buttons

Dierk König Dierk.Koenig@canoo.com
Fri, 1 Mar 2002 12:09:58 +0100


sorry, I accidentially sent the mail to marcus only, not to the list.

-----Original Message-----
From: Dierk König [mailto:Dierk.Koenig@canoo.com]
Sent: Mittwoch, 27. Februar 2002 12:30
To: Marcus Ahnve
Subject: RE: [Webtest] Forms and image buttons


Hi Marcus,

> -----Original Message-----
> From: Marcus Ahnve [mailto:marcus.ahnve@lecando.com]

> Thanks for your reply. I must have been very stressed when I wrote this
> yesterday because I meant _frames_ instead of forms. Sorry about that -
> the question became really ridiculous. However, what about frames?

Frames in short:
- not supported.
- not intended to be supported.
- best practice: "invoke" the pages that get called by <frame src="page">
directly.

Here some reasoning:

I would be easy to detect that a page is a frameset.
However, verification could be applied to only one of the "src" pages.
Selection of that page is no different than just invoking it.

A common logic that someone may want to test is when clicking
a link in one frame changes the appearance of a different frame
of the same framset.
Fault model:
	target misssspelled,
	wrong target chosen,
	improper frameset nesting,
	"forceToUpperFrame"-handling,
	call by external referers creates nested framesets,
	external links have unexpected behavior of frameset nesting
	(more?)
Test model:
	verify anchor elements against expected target patterns
	introduce patterns tests for every defect found (through users or manual
tests)
	verify anchor elements against absence of defect patterns (<NOT>-step)

--- you may want to stop reading here ---

WebApps based on frames are a little harder and less reliable to test.

Some testing specialists like Frank Westphal claim that every test problem
is an architecture or design problem.

I'm not surprised seeing less and less of the "big" sites on the web using
frames
(JavaScript, Applets, ActiveX, Flash, ...).

cheers
Dierk