For a long time i have talked about how important it is for Watir to make
code like these examples work:
@ie_new = Watir::IE.new
text_field = @ie_new.text_field(:name, 'text1')
@ie_new.goto($htmlRoot + "textfields1.html")
assert_equal('Hello World', text_field.value)
$ie.goto($htmlRoot + "textfields1.html")
text_field = $ie.text_field(:name, 'text1')
$ie.refresh
assert_equal('Hello World', text_field.value)
Currently, these both have errors, in the first place, you get an ugly OLE
error, in the second, the reference is stale and text_field.value returns
an empty string. This is something that often surprises and confuses new users.
I haven't really know what to call the ability to make these examples work.
For a while i called it mapping, because it allows you to create predefined
window maps, if you want to, but that got Jonathan all prickly because of
years he'd spent suffering from having to update SilkTest window maps.
You could also call it lazy instantiation. Not quite accurate, but it gives
the right impression. Or late binding.
At the moment i think i want to call it deferred element binding, because
the Watir object wouldn't actually get mapped to an element in a page until
as late as possible. This would really loosen the binding between an
individual Watir object and the page elements it can refer to -- and now a
single Water object can in fact be used to refer to the same element on
different pages or simply successive page loads (which previously all
required new Watir objects). And it would also loosen the binding between
Watir and COM, making it easier to support bindings to other browser
driving technologies, including Selenium and XP/COM. In other words:
Firefox support.
Actually, i lied. The examples above do in fact work with the latest
version of Watir, found in HEAD. I just wrote the needed code tonight.
Right now it only works with text boxes. And only in the simple case where
it is in an IE container. But i've refactored enough of the code that the
path ahead to making deferred element binding work with all elements and
all containers is simply a matter of continuing the same kinds of
refactorings that i've been making over the past few months anyhow. By
removing duplication and reducing the size of the Watir code base, i am
sure that we can get this feature working across the board.
Particular thanks go to Elisabeth Hendrickson who prodded me to prove to
her that this would be hard. In the process i realized it wouldn't be quite
so hard after all.
Bret
_____________________
Bret Pettichord
www.pettichord.com
_______________________________________________
Wtr-general mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/wtr-general
