Well, that's exactly what is failing in my tests currently:
public boolean processChildren() {
IFormSubmitter submitter = this.getFormSubmitter(); //!\\
return submitter != null && submitter.getForm() == this;
}
If this is the way to control submit to nested forms, we'll have to make
sure this works in tests too.
Sven
On 01/30/2013 12:16 PM, Martin Grigorov wrote:
Recently Sebastien (the author of wicket-jquery-ui) needed to set the
submitting component from the client side (JS API).
He asked about wicketSubmitFormById(), i.e. Wicket 1.5 API but I guess he
will need the same in Wicket 6.
http://markmail.org/thread/2onx3qrq5l3vrsyc
Without the info what is the submitting button you wont be able to decide
whether an outer or nested form is submitted.
On Wed, Jan 30, 2013 at 12:04 PM, Sven Meier <[email protected]> wrote:
it will always see AFSB's anonymous IFormSubmitter as submitter
Yes.
and wont ever know that AjaxSubmitLink is the real submitter behind
I'm not sure why anybody should be interested in this detail:
If an AjaxFallbackButton is triggered via Ajax, who is the submitter, the
button or the behavior?
IMHO it's the behavior.
I think this is the broken code
I think we should just remove/deprecate ARA#submittingComponentName ... it
works fine without it.
Sven
On 01/30/2013 10:31 AM, Martin Grigorov wrote:
I see what happens.
In Form#onFormSubmitted there is:
if (submitter == null)
{
submitter = findSubmittingButton();
}
i.e. it will always see AFSB's anonymous IFormSubmitter as submitter and
wont ever know that AjaxSubmitLink is the real submitter behind.
I think this is the broken code.
I'm not sure how important is to know that a link is the actual submitter
though. AFSB's anonymous IFormSubmitter will delegate to the link anyway.
On Wed, Jan 30, 2013 at 10:13 AM, Sven Meier <[email protected]> wrote:
The AFSB knows that it is triggered, because it is directly triggered as
a
behavior listener.
For that it doesn't need the extra parameter.
My concern is that Form#findSubmittingButton() is inconsistent between
runtime and tests.
Before digging into WicketTester I'd rather remove that paramter in ARA.
Sven
On 01/30/2013 09:27 AM, Martin Grigorov wrote:
It is used indirectly by AjaxSubmitLink and AjaxButton, for example.
They use AFSB. After submit the form (and the client code) know that a
link/button actually submitted the form.
Or maybe I don't understand your concern.
On Wed, Jan 30, 2013 at 9:19 AM, Sven Meier <[email protected]> wrote:
I know, but for what is that needed?
Sven
On 01/30/2013 09:09 AM, Martin Grigorov wrote:
Hi Sven,
AFSB sets it only if :
if (getComponent() instanceof IFormSubmittingComponent)
{
String submittingComponentName =
((IFormSubmittingComponent)******getComponent()).getInputName(**
)****;
attributes.******setSubmittingComponentName(******
submittingComponentName);
}
i.e. if it is attached to a submitting component like a button.
It is the same in 1.5
(org.apache.wicket.ajax.form.******AjaxFormSubmitBehavior#****
getEventHandler):
if (getComponent() instanceof IFormSubmittingComponent)
{
call.append("'")
.append(((******IFormSubmittingComponent)****
getComponent()).getInputName()*****
*)
.append("' ");
}
On Tue, Jan 29, 2013 at 11:37 PM, Sven Meier <[email protected]> wrote:
Hi,
can somebody explain what AjaxRequestAttributes#****
setSubmittingComponentName()
is for?
It is set by AjaxFormSubmitBehavior#********updateAjaxAttributes()
only.
Why I'm asking?
When a form is submitted by an AjaxFormSubmitBehavior, we have an
inconsistency between WicketTester and a real submit from the
browser:
In the former case Form#findSubmittingButton() returns null, in the
latter
case it returns the Button the AjaxFormSubmitBehavior is bound to.
It seems that #setSubmittingComponentName() is superfluous because
the
AjaxFormSubmitBehavior will be triggered as the form submitter
anyway.
Strictly speaking the button is *not* the form submitter but the
behavior
is, so it makes sense that #findSubmittingButton() should return
null.
Any thoughts?
Sven
