4 messages in com.googlegroups.pylons-discussRe: problems with observe_field when ...| From | Sent On | Attachments |
|---|---|---|
| Eric Ongerth | 24 Feb 2008 15:48 | |
| Eric Ongerth | 24 Feb 2008 16:03 | |
| Eric Ongerth | 25 Feb 2008 22:11 | |
| Eric Ongerth | 26 Feb 2008 00:32 |
| Subject: | Re: problems with observe_field when form is inside of an update_element_function |
|---|---|
| From: | Eric Ongerth (eric...@public.gmane.org) |
| Date: | 02/24/2008 04:03:44 PM |
| List: | com.googlegroups.pylons-discuss |
Update. Usually I buffer all my <%def>s because I'm almost always using them functionally in places where the output needs to be held until it's time to insert it. In this case, however, I tried unbuffering the trial() def and pow, the Ajax fires like it should.
..sort of. Actually, it's very erratic and I'm wondering why. In my real application, there's not just the two test fields, but six fields, only one of which is observed. I click the button, the form appears. Now if I enter text for each field starting with the first field, and progressing from one field to the next via the Tab key, the observer call fires as soon as I Tab out of the observed field. So far so good. But it does not fire if I leave the field via shift- Tab. It does not fire if I mouse-click straight into that observed fourth field and fill it and then Tab forward out of it. Basically it *only* fires if I fill in the form like a good little boy, starting with the first field and putting something in every field, never going back and never jumping around. How useful is that?
So there must still be something I'm missing. I've got my _with property set properly, so when the call does get made, the POST header is coming through correctly. I've tried frequency=0, frequency= various positive numbers between 0 and 1. What is causing the inconsistent behavior?
Also, sometimes the Ajax call takes a good two seconds or more to fire, and my system is not under any sort of load right now (this is all being done on the home workstation with very little network activity). Sometimes several of the calls seem to get backed up and then they all fire in rapid sequence after a delay of several seconds. This is not so great. The purpose of observing this field is, if the user enters certain values, two more fields need to be added to the form. If there's that much delay it's definitely not the smoothness for which I sought Ajax. Should I try changing this one operation to synchronous?
More thanks. Eric
On Feb 24, 3:48 pm, Eric Ongerth
<eric...@public.gmane.org> wrote:
Two cases... one works, one does not, and the difference seems to be that in Case 2 (non-working) the form is displayed in the view only after the user clicks a button that triggers an update_element_function to show the form. So there seems to be something I'm not understanding -- maybe about filtering or escaping? -- which is preventing the observer from working. Too bad no error messages in the non-working case, just silence where I expect an Ajax call. The following are in a Mako template with the standard pylons helpers %included at the top.
<%def name="trial()" buffered="True"> ${h.form_remote_tag( url=h.url_for( controller="FVal", action="add", feature_id=999, parentFVal_id=666), complete=h.evaluate_remote_response(), method="POST")} test feature main: ${h.text_field(name="main value")} ${h.text_field(name="foo")} % if True: ${h.observe_field( "foo", with_="{foo:value}", url=h.url_for( controller="FVal", action="alter_input_form"))} % endif ${h.submit("save", name="commit")} ${h.end_form()} </%def>
this version WORKS: ${trial()}
but this one DOES NOT WORK: (no Ajax call ever happens) <div id="whatever"> ${h.button_to_function( "Test", h.update_element_function( "whatever", action="update", content=(trial())))} </div>
I'm looking at the results via Firebug. As far as I can tell, both versions generate the same HTML and same javascript. The only difference I can see is that in the 2nd case, the form is not put on the page until the user clicks the "Test" button.
Why should the update_element_function somehow result in breaking the Ajax of the observe_field function? The form in each case is identical:
<form onsubmit="new Ajax.Request('/FVal/add? feature_id=999&parentFVal_id=666', {asynchronous:true, evalScripts:true, method:'POST', onComplete:function(request) {eval(request.responseText)}, parameters:Form.serialize(this)}); return false;" method="post" action="/FVal/add? feature_id=999&parentFVal_id=666"> test feature main: <input type="text" name="main value" id="main value"/> <input type="text" name="foo" id="foo"/> <script type="text/javascript"> //<![CDATA[ new Form.Element.EventObserver('foo', function(element, value) {new Ajax.Request('/FVal/alter_input_form', {asynchronous:true, evalScripts:true, parameters:{foo:value}})}) //]]> </script> <input type="submit" value="save" name="commit"/> </form>
And specifically, the Form.Element.EventObserver generated in both cases is identical.
So there's gotta be something I'm missing! Any insight would be most welcome, thanks.
Eric