Wiki Markup |
---|
{scrollbar} |
Ajax Components
Main article: Ajax and Zones
Contents
Table of Contents | ||||
---|---|---|---|---|
|
Do I have to specify both id
and t:id
for Zone components?
The examples for the Zone component (in the Component Reference) consistently specify both id
and t:id
and this is probably a good idea.
...
The point is, to be sure, specify the exact client id. This will be the value for the zone
parameter of the triggering component (such as a Form, PageLink, ActionLink, etc.).
How do I update the content of a Zone from an event handler method?
When a client-side link or form triggers an update, the return value from the event handler method is used to construct a partial page response; this partial page response includes markup content that is used to update the Zone's client-side <div>
element.
Where does that content come from? You inject it into your page.
Code Block | ||||
---|---|---|---|---|
| ||||
<t:zone id="search" t:id="searchZone">
<t:form t:id="searchForm" zone="searchZone">
<t:textfield t:id="query" size="20"/>
<input type="submit" value="Search"/>
</t:form>
</t:zone>
<t:block id="searchResults">
<ul>
<li t:type="loop" source="searchHits" value="searchHit">${searchHit}</li>
</ul>
</t:block>
|
Code Block | ||||
---|---|---|---|---|
| ||||
@Inject
private Block searchResults;
Object onSuccessFromSearchForm()
{
searchHits = searchService.performSearch(query);
return searchResults;
}
|
So, when the search form is submitted, the resulting search hits are collected. In the same request, the searchResults block is rendered, package, and sent to the client. The form inside the client-side Zone <div>
is replaced with the list of hits.
In many cases, you just want to re-render the Zone itself, to display updated content. In that case, you don't need a separate <t:block>
, instead you can use @InjectComponent to inject the Zone object itself, and return the Zone's body:
Code Block | ||||
---|---|---|---|---|
| ||||
@InjectComponent private Zone statusZone; Object onActionFromUpdateStatus() { return statusZone.getBody(); } |
How to I update multiple zones in a single event handler?
To do this, you must know, on the server, the client ids of each Zone. That's one of the reasons that you will generally set the Zone's client id (via the Zone.'s id parameter), rather than let Tapestry assign a client id for you.
From the event handler method, instead Instead of returning a Block or a Component, return a multi-zone update:
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
@Inject private Block searchResults; @Inject private Block statusBlock; Object@Inject private AjaxResponseRenderer ajaxResponseRenderer; void onSuccessFromSearchForm() { searchHits = searchService.performSearch(query); message = String.format("Found %,d matching documents", searchHits.size()); ajaxResponseRenderer.addRender("results", searchResults).addRender("status", statusBlock); return new } |
Note: Users of Tapestry 5.2 and earlier (which didn't support AjaxResponseRenderer) must replace that last line with: return new MultiZoneUpdate("results",
...
searchResults).add("status",
...
statusBlock);
AjaxResponseRenderer adds other useful commands as well. It also has the advantage that a simple return value can be returned to render content for the Zone that triggered the request.
What's that weird number in the middle of the client ids after a Zone is updated?
You might start with markup in your template for a component such as a TextField:
Code Block | ||||
---|---|---|---|---|
| ||||
<t:textfield t:id="firstName"/>
|
When the component initially renders as part of a full page render, you get a sensible bit of markup:
Code Block | ||||
---|---|---|---|---|
| ||||
<input id="firstName" name="firstName" type="text">
|
But when the form is inside a Zone and rendered as part of a zone update, the ids get weird:
Code Block | ||||
---|---|---|---|---|
| ||||
<input id="firstName_12a820cc40e" name="firstName" } type="text"> |
What's happening here is that Tapestry is working to prevent unwanted id clashes as part of the page update. In an HTML document, each id
is expected to be unique; most JavaScript is keyed off of the id
field, for instance.
In a full page render, components don't just use their component id (t:id
) as their client id; instead they use the JavaScriptSupport
environmental to allocate a unique id. When there's no loops or conflicts, the client id matches the component id.
When the component is inside a loop, a suffix is appended: firstName
, firstName_0
, firstName_1
, etc.
When the component is rendered as part of an Ajax partial page update, the rules are different. Since Tapestry doesn't know what content has been rendered onto the page previously, it can't use its normal tricks to ensure that ids are unique.
Instead, Tapestry creates a random-ish unique id suffix, such as "12a820cc40e" in the example; this suffix is appended to all allocated ids to ensure that they do not conflict with previously rendered ids.
Why do I sometimes get the exception "The rendered content did not include any elements that allow for the positioning of the hidden form field's element." when rendering an empty Zone?
As part of Tapestry's form processing, it must write a hidden input element with information needed when the form is submitted. Since the content of a Zone may be changed or removed, a hidden field is created just for the Zone, separate from the rest of the enclosing form.
At the same time, Tapestry wants to position the <input> field in a valid location, and HTML defines some constraints for that; an input field must appear inside a <p> or <div> element. If your zone is initially empty, there's no place to put the hidden element, and Tapestry will complain.
The solution is simple: just add a <div> element to the body of the zone. This ensures that there's a place for the hidden input field. An empty <div> element (even one containing a hidden form field) will not affect page layout.
Wiki Markup |
---|
{scrollbar} |