This little OpenLayers.Control.EditingToolbar from is inserted by default:
It's not really evident what these buttons mean. I would like to replace this editing toolbar with a button group (e.g. like the one Twitter Bootstrap offers):
The markup of the editing toolbar currently is this:
<div id="panel" class="olControlEditingToolbar">
<div class="olControlNavigationItemInactive olButton"></div>
<div class="olControlDrawFeaturePointItemActive olButton"></div>
<div class="olControlDrawFeaturePathItemInactive olButton"></div>
<div class="olControlDrawFeaturePolygonItemInactive olButton"></div>
</div>
The images are basic sprites – so I know I could change these. But I can't see how I could get away from these divs, replacing them with buttons. I thought about just creating the button group manually and add click() event listeners to the buttons, triggering OpenLayers' different editing modes. But I couldn't find any documentation on how I could do that.
So, basically, I see these options:
Create button group manually, and trigger the appropriate OpenLayers events through my own JS — but which events do I need to trigger?
Don't use the EditingToolbar, but manually build my toolbar with OpenLayers — how could I do that?
Modify the automatically created editing toolbar by hacking the OpenLayers source (meh…) — is this worth the effort?
The best way is to manually build the control buttons. Based on the Draw Feature Example, you can go ahead and add your controls:
drawControls = {
point: new OpenLayers.Control.DrawFeature(pointLayer,
OpenLayers.Handler.Point),
line: new OpenLayers.Control.DrawFeature(lineLayer,
OpenLayers.Handler.Path),
polygon: new OpenLayers.Control.DrawFeature(polygonLayer,
OpenLayers.Handler.Polygon),
)
};
for(var key in drawControls) {
map.addControl(drawControls[key]);
}
Then, add a function that changes the currently used control based on a clicked element:
function toggleControl(element) {
for(key in drawControls) {
var control = drawControls[key];
if(element.value == key && element.checked) {
control.activate();
} else {
control.deactivate();
}
}
}
Finally, create the HTML markup yourself. For each element that changes the control, add an onClick handler that calls the toggleControl function. You can also attach the click handler through jQuery, but in essence, this works:
<ul id="controlToggle">
<li>
<input type="radio" name="type" value="none" id="noneToggle"
onclick="toggleControl(this);" checked="checked" />
<label for="noneToggle">navigate</label>
</li>
<li>
<input type="radio" name="type" value="point" id="pointToggle" onclick="toggleControl(this);" />
<label for="pointToggle">draw point</label>
</li>
<!-- add more elements here, based on which draw modes you added -->
</ul>
You can see this in action here (sign up with a test user, no real e-mail needed) and find the code on GitHub.
Related
In my Office add-in I have a checkbox like the following:
<div class="ms-CheckBox">
<input id="inputId" type="checkbox" class="ms-CheckBox-input" />
<label id="labelId" role="checkbox" class="ms-CheckBox-field" aria-checked="false" name="checkboxA" for="inputId>
<span class="ms-Label">Text</span>
</label>
</div>
I want to retrieve through JavaScript its checked status (or its aria-ckecked status, I'm still not getting the differences between them), which I thought was through document.getElementById( 'labelId' ).checked, since it's specified in the documentation that they have an optional checked member, but I only get an undefined with it.
I'm very new to these technologies and have a couple concerns:
Does "optional member" mean that I have to explicitly create it so that it exists? If so, how can I do that?
However the checked member may come to existance, do I have to manually handle its value every time it's clicked on by the user or is it already internally managed and I simply haven't found the way to access it yet?
Maybe I just can't see a mistake I've made on the html code for the checkbox?
Thank you in advance!
You have several sources of documentation on Office UI Fabric depend on framework you are using or about to use. Your choices are:
JavaScript only (no framework)
React
Angular
Form the look up table you would choose JavaScript only link and follow it to find the component you are interested in. Before that I would suggest to read "Get Started using Fabric JS".
Now when you have documentation on checkbox component of vanilla JS implementation, follow the steps to set up your checkbox. This would include:
Confirm that you have references to Fabric's CSS and JavaScript on your page
Copy the HTML from one of the samples below into your page.
<div class="ms-CheckBox">
<input tabindex="-1" type="checkbox" class="ms-CheckBox-input">
<label role="checkbox" class="ms-CheckBox-field" tabindex="0" aria-checked="false" name="checkboxa">
<span class="ms-Label">Checkbox</span>
</label>
</div>
Add the following tag to your page, below the references to Fabric's JS, to instantiate all CheckBox components on the page.
<script type="text/javascript">
var CheckBoxElements = document.querySelectorAll(".ms-CheckBox");
for (var i = 0; i < CheckBoxElements.length; i++) {
new fabric['CheckBox'](CheckBoxElements[i]);
}
</script>
To get the status of your checkbox use method getValue() which returns true or false whether the component is checked or not.
usually I can figure out a way to make Knockout-js do what I want. In this case however, i'm struggling a little, so I'm opening the question up to the community here on SO.
Introduction
I'm writing an HTML5 web app using typescript, bootstrap, knockoutjs and a nodejs backend.
In my UI which is all controlled via knockoutJS I have a set of buttons, formed as a bootstrap 3 button group of select-able options.
This justified group, gives me 4 buttons horizontally, but allows the behaviour of the button selections to remain consistant with a group of option buttons.
That consistancy is important, beacuse ONLY one button at a time can ever be selected, so when one is clicked, the rest deselect.
This is a default component in BS3, as the following image shows:
As you can see in the image, the 'Discarded' button is selected, to achieve this a class of 'active' must be added to the existing class list of the label element surrounding the inner radio element that makes up the control. The following HTML is used to create this image:
<div class="btn-group btn-group-justified" data-toggle="buttons">
<label class="btn btn-primary">
<input type="radio" name="options" id="option1" checked >Rejected
</label>
<label class="btn btn-primary active">
<input type="radio" name="options" id="option2">Discarded
</label>
<label class="btn btn-primary">
<input type="radio" name="options" id="option3">Held
</label>
<label class="btn btn-primary">
<input type="radio" name="options" id="option3">Header Added
</label>
</div>
All this works great except for one small flaw, I'm using knockout JS to manage the UI.
The Problem I'm Trying to Solve
The checked state of each of the options is tied to a property inside the view model applied to the HTML, so the inner option on the rejected button for example has a normal knockout-js checked binding added to it as follows:
<input type="radio" name="options" id="option1" checked data-bind="checked: reject">Rejected
Each of the options, each have their own backing field:
reject
discard
hold
addheader
and each of those backing fields are a standard boolean value holding true/false, what I can't figure out is how to add/remove the 'active' class on the enclosing label, to reflect which of these states has been selected.
To be more precise, I cant figure out the best way to do it elegantly.
Approaches I've tried
what I know works is to add a simple computed observable to each label that returns
"btn btn-primary active"
when that option is set to true, and
"btn btn-primary"
when it is not.
I know this, because in my view model, I had a simple function:
SenderDialogViewModel.prototype.isRejectSelected = function () {
if (this.reject == true) {
return "btn btn-primary active";
}
return "btn btn-primary";
};
However, this approach means 4 functions, one for each flag to test, making it difficult to add new flags at a later date.
What I'd like to be able to do, is something like the following:
<label class="btn btn-primary" data-bind="class: isSelected(reject)">
as an example.
Which I almost got to work with a slight modification to the above:
SenderDialogViewModel.prototype.isSelected = function (selectionVariable) {
if (selectionVariable == true) {
return "active";
}
return "";
};
Where selection variable could be any of the flags available in the view model, passed in.
The problem here was, that this ONLY updated the first time the UI was drawn, subsequent changes to the flags, failed to update the UI to reflect the given status.
To try and resolve this, I changed the function to a computed observable, only to then receive a JS error when the UI was drawn, stating that the computed observable had to have a 'write' handler added to it, because I was passing a parameter in.
If I need to add a write handler, then that's fine, but I'd rather not.
Summary
So in summary, there are ways of changing the class list in sync with other options, but most of them are messy, what I'm trying to do is create a way that's easily expanded as new buttons are added (This is important as some button sets are dynamically generated), rather than adding a handler to individually check and report the status on each and every variable there, in one function call that can be added simply into the view-model and re-used again and again.
Ok... and as it ALWAYS happens, no sooner do I post this, than I actually figure out how to make it work.
The solution was staring me in the face all along, in the form of the knockout js css binding.
To quote the knockout-js docs:
The css binding adds or removes one or more named CSS classes to the associated DOM element. This is useful, for example, to highlight some value in red if it becomes negative.
What this is saying, as "I can apply or remove a single class, to the collection of classes already present, based on the value of a variable in my view model"
So, the answer to my problem, quite simply becomes:
<div class="btn-group btn-group-justified" data-toggle="buttons">
<label class="btn btn-primary" data-bind="css: { active: reject }">
<input type="radio" name="options" id="option1" checked >Rejected
</label>
<label class="btn btn-primary" data-bind="css: { active: discard }">
<input type="radio" name="options" id="option2">Discarded
</label>
<label class="btn btn-primary" data-bind="css: { active: hold }">
<input type="radio" name="options" id="option3">Held
</label>
<label class="btn btn-primary" data-bind="css: { active: addheader }">
<input type="radio" name="options" id="option3">Header Added
</label>
</div>
Along with that, if I now just add the appropriate option/checked bindings to the option controls themselves, then everything should update correctly as needed when the buttons are clicked.
A little side note on working your own answer out
I think sometimes, the exercise of just having to think through your problem, while 'virtually' trying to describe it to others, triggers the brain to think in a different direction.
It certainly helped, that I was typing this when the penny dropped, but I proceeded and then answered/updated as appropriate, because it occurs to me that this is a question that will trip others up too, hopefully this will serve as an education to fellow travelers who might like me just be suffering from a sunday evening brainfart.
Update Monday 30-June 2014
It turns out, this was a little more tricky than I first anticipated. Sure I solved the main answer to my question about syncing the CSS for the button above. BUT... syncing the Option buttons also turned out to be quite a challenge, this update is to present a full end to end solution.
First, you need to mark up your HTML like this:
<p><strong>Rule type:</strong></p>
<div class="btn-group btn-group-justified" data-toggle="buttons">
<label class="btn btn-primary" data-bind="css: { active: reject }, click: function(){ changeType('reject'); }">
<input type="radio" name="ruletype" value="reject" data-bind="checked: selectedOptionString" >Rejected
</label>
<label class="btn btn-primary" data-bind="css: { active: discard }, click: function(){ changeType('discard'); }">
<input type="radio" name="ruletype" value="discard" value="true" data-bind="checked: selectedOptionString">Discarded
</label>
<label class="btn btn-primary" data-bind="css: { active: hold }, click: function(){ changeType('hold'); }">
<input type="radio" name="ruletype" value="hold" data-bind="checked: selectedOptionString">Held
</label>
<label class="btn btn-primary" data-bind="css: { active: addheader }, click: function(){ changeType('addheader'); }">
<input type="radio" name="ruletype" value="addheader" data-bind="checked: selectedOptionString">Header Added
</label>
</div>
The KEY take away's here are as follows:
1) The CSS rule must specify the class 'active' and be tied to your independent option flag that shows true/false for that option being selected.
2) You MUST have the click handler on the button, BS3 (as I found out) processes the click on the button NOT on the option control, due to how knockout works this HAS to be an inline function, passing in a single parameter, DO NOT be tempted to tie it directly to the computed observable used by the option, it won't work.
3) You must mark the option elements up as shown, that is they must ALL have the same name attribute, the value MUST match what you want that selected option to portray, and must match the strings your button handler is sending
4) Each option element cannot be bound to a simple variable, you need to pass it through a computed observable, not just because of how I'm handling them, but even for simple single Boolean switches it uses "true" & "false" as strings and not as Boolean's as you might expect it.
Once you've marked up your HTML, you then need to build a view model to support it all, in my case I actually did this using typescript then compiled to JS, the code I'm pasting in here is the JS code produced by the TS compiler.
first and foremost you need to make sure you have the following properties on your view model:
this.reject = ko.observable(false);
this.discard = ko.observable(false);
this.hold = ko.observable(false);
this.addheader = ko.observable(false);
(use self, this, me... or what ever it is you use to define your knockout models) the important thing is that they are simple ko observable boolean's
You also need a computed observable that has both a write and a read function:
this.selectedOptionString = ko.computed({
read: function () {
if (this.reject())
return "reject";
if (this.discard())
return "discard";
if (this.hold())
return "hold";
if (this.addheader())
return "addheader";
},
write: function (value) {
console.log("rejectstr:");
console.log(value);
if (value == "reject") {
this.reject(true);
this.discard(false);
this.hold(false);
this.addheader(false);
}
if (value == "discard") {
this.reject(false);
this.discard(true);
this.hold(false);
this.addheader(false);
}
if (value == "hold") {
this.reject(false);
this.discard(false);
this.hold(true);
this.addheader(false);
}
if (value == "addheader") {
this.reject(false);
this.discard(false);
this.hold(false);
this.addheader(true);
}
}
}, this);
This could probably be done a lot more elegantly, but essentially when an option in a group is activated under knockout, knockout takes whatever is in the 'Value' attribute and sends that as a string into your view model.
If you tied this to a simple observable, then that observable would get set to the string value of that attribute. For me however, because I have a series of 4 flags to set that control various states on the UI, a chained if then was appropriate (a switch or possibly a lookup array or hashtable would have worked just as well)
The ultimate outcome of the observable is that one boolean and one only ever be set at a time, and beacuse the CSS in the button is tied to this flag, then the active class gets applied to the given button for which ever is set to true.
For a read, you need to translate your flag state back to a string for knockout to compare to the values it knows about, so the read does the reverse.
For the button click handler, you have to do this inline as shown in the markup, this is beacuse knockout reserves some parameters for automatic things like element name, event info and other's, none of which we need here, so the simplest way is to inline it.
However, in lining it means your not tying to a property and so you can't tie it directly to the computed observable used by the option controls.
Instead what you need to do is add a small stub function to your view model as follows:
SenderDialogViewModel.prototype.changeType = function (newType) {
this.selectedOptionString(newType);
};
This does not need to be observable in any way as it only gets called one way when your button is clicked.
If you want to see the full solution, I'll be making the app this is part of available on my git-hub page free for people to use, but it's not finished yet.
Hope everything above however turns out to be useful for some folk, I have to admit, it turned out to be a little bit more of a challenge than I expected.
Shawty
To get an idea of the setup I’m using in my application I set up this simplified example:
<div ng-controller="Ctrl">
<ul>
<li ng-repeat="oConfigEntry in oConfiguration.oConfigEntriesColl">
<ul>{{oConfigEntry.sDescription}}
<li ng-repeat="oConfigSubEntry in oConfigEntry.oConfigSubEntriesColl">{{oConfigSubEntry.sDescription}}
<input type='checkbox' ng-model='oConfigSubEntry.bNoOption' />{{oConfigSubEntry.bNoOption}}
<ul>
<li ng-repeat='oConfigSubSubEntry in oConfigSubEntry.oConfigSubSubEntriesColl'>{{oConfigSubSubEntry.sDescription}}
<input type='number' placeholder='length' ng-model='oConfigSubSubEntry.dLength' />
<input type='number' placeholder='width' ng-model='oConfigSubSubEntry.dWidth' />
<input type='number' placeholder='height' ng-model='oConfigSubSubEntry.dHeight' />
<input type='checkbox' title='opt1' ng-model='oConfigSubSubEntry.bOpt1' />
<input type='checkbox' title='opt2' ng-model='oConfigSubSubEntry.bOpt2' />
</li>
</ul>
</li>
</ul>
</li>
</ul>
<pre ng-bind="oConfiguration | json"></pre>
</div>
see http://jsfiddle.net/ppellegr/4QABQ/
Unfortunately the problem I’m facing in the real application cannot be reproduced in the latter mentioned example.
The problem is that in the real application the checkboxes are not clickable. Clicking the checkboxes do not check them. The checkboxes remain unchecked.
The other way around If the corresponding model is initialized the checkboxes are checked but cannot be unchecked by clicking them.
Even plain checkboxes with no model assigned cannot be checked if they are placed within a nested ng-repeat.
e.g.
<input type="checkbox" />
Has anyone already noticed such a phenomenon?
additional observations:
The first click on the checkbox changes the value of the model.
Subsequent clicks do not change the value. The value of the model remains the
same.
While the first click on the checkbox changes the value of the
model, the checkbox itself remains checked/unchecked depending on the
inital value of the model.
My guess is that another element is positioned in such a way as to cover or overlap the checkbox, and that is preventing you from interacting with it. Assuming you have no inline styles applied to your markup, you can test this easily by disabling CSS in your browser (you may need to install an extension to do this, eg: How to disable CSS in Browser for testing purposes).
If you find that you can click the checkbox like that, you then need to debug your css to find the offending element. Use firebug or chrome developer tools to explore the markup and css.
In the real application the checkbox showing the described behavior is within a list that is enriched by a little jQuery feature making the list collapsible and expandable...
function prepareList() {
$('#ConfigContainer').find('li:has(ul)')
.click(function (event) {
if (this == event.target) {
$(this).toggleClass('expanded');
$(this).children('ul').toggle('medium');
}
return false;
})
.addClass('collapsed')
.children('ul').hide();
}
$(document).ready(function () {
prepareList();
});
see http://jsfiddle.net/ppellegr/cP726/
In this example the described behavior can be reproduced...
The culprit line of code is obvious:
return false;
This stops propagation of event and obviously interferes with the checkbox...
Lesson learnt:
check whether javascript, jQuery or the like are interfering with angularjs...
consider writing an angular directive...
Is there a specific reason that most everyone implements edit-in-place as a shown 'display' div and a hidden 'edit' div that are toggled on and off when somebody clicks on the associated 'edit' button like so?
<div id="title">
<div class="display">
<h1>
My Title
</h1>
</div>
<div class="edit">
<input type="text" value="My Title" />
<span class="save_edit_button"></span>
Cancel
</div>
</div>
Everywhere I look, I see edit-in-place basically handled like this. This approach certainly makes sense when you are rendering all views on the server side and delivering them to the client. However, with pure AJAX apps and frameworks like backbone.js, it seems that we could make our code much more DRY by rendering edit-in-place form elements on the fly as necessary, possibly even making a factory method that determines which form element to render. e.g.
an H1 element with class "title" is replaced by <input type="text" />
a span with class "year_founded" is replaced by <input type="number" min="1900" max="2050" />
a span with class "price" is replaced by an input with the appropriate mask to only allow prices to be input.
Is this practice of rendering all edit-in-place form elements a historical legacy leftover from when pages were rendered on the server-side?
Given the flexibility and power we have with client-side MVC frameworks like Backbone.js, is there a reason for not creating and inserting the form elements on the fly when necessary using a factory method? Something like this:
HTML
<div id="description">
Lorem ipsum dolar set amit...
</div>
<span class="edit_button"></span>
Backbone.js View
events: {
"click .edit_button": "renderEditInPlaceForm",
},
renderEditInPlaceForm: function:(e) {
var el = $(e.currentTarget).previous();
var id = el.attr('id');
var value = el.text();
var tagName = el.tagName();
var view = new editInPlaceForm({
id: id,
type: tagName,
value: value
});
$("#id").html(view.render().el)
},
Where editInPlaceForm is a factory that returns the appropriate edit-in-place form element type based on tagName. This factory view also controls all its own logic for saving an edit, canceling an edit, making requests to the server and rerendering the appropriate original element that was replaced with the .html() function?
It seems to me that if we use this approach then we could also render the <span class="edit_button"></span> buttons on the fly based on a user's editing rights like so:
<h1 id="title">
<%= document.get("title") %>
</h1>
<% if (user.allowedToEdit( document, title )) { %>
<span class="edit_glyph"></span>
<% } %>
where the allowedToEdit function on the user model accepts a model and attribute as its arguments.
It's an interesting idea. The devil is in the detail.
While your simple example is easily rendered as an editable form on the fly, things quickly get trickier when dealing with other data types.
For example - suppose my edit form requires the user to choose a value from a select list. On the display form I can simply display the user's choice, but for the edit form I am going to need those other available choices. Where do I hide them on the display? Similar issues exist for checkboxes, radio lists...
So, perhaps we should consider rendering the edit form, and then deriving our display-view from that?
After 5 Backbone apps I came to same thoughts.
When things are complicated you have forms to show relations between user data,
but in simple cases you just need input, select, checkbox over h1, div or span
Now I am searching for jQuery plugin to make simple in place editing without ajax.
jQuery but not Backbone becuase I don't want to be tight coupled with Backbone for such small thing.
Likely to wright my own jQuery + Synapse plugin http://bruth.github.com/synapse/docs/.
Synapse for binding with model and jQuery for input placing
I'm making a website using JSP and servlets and I have to now break up a list of radio buttons to insert a textarea and a button. I have got the button and textarea to hide and show when you click on the radio button it shows the text area and button. But this only appears at the top and when there are hundreds on the page this will become awkward so i need a way for it to appear underneath. Here is what my HTML looks like when complied:
<form action="addSpotlight" method="POST">
<table>
<tr><td><input type="radio" value="29" name="publicationIDs" ></td><td>A System For Dynamic Server Allocation in Application Server Clusters, IEEE International Symposium on Parallel and Distributed Processsing with Applications, 2008</td> </tr>
<tr><td><input type="radio" value="30" name="publicationIDs" ></td><td>Analysing BitTorrent's Seeding Strategies, 7th IEEE/IFIP International Conference on Embedded and Ubiquitous Computing (EUC-09), 2009</td> </tr>
<tr><td><input type="radio" value="31" name="publicationIDs" ></td><td>The Effect of Server Reallocation Time in Dynamic Resource Allocation, UK Performance Engineering Workshop 2009, 2009</td> </tr>
<tr><td><input type="radio" value="32" name="publicationIDs" ></td><td>idk, hello, 1992</td> </tr>
<tr><td><input type="radio" value="33" name="publicationIDs" ></td><td>sad, safg, 1992</td> </tr>
<div class="abstractWriteup"><textarea name="abstract"></textarea>
<input type="submit" value="Add Spotlight"></div>
</table>
</form>
Now here is what my JSP looks like:
<form action="addSpotlight" method="POST">
<table>
<%int i = 0; while(i<ids.size()){%>
<tr><td><input type="radio" value="<%=ids.get(i)%>" name="publicationIDs" ></td><td><%=info.get(i)%></td> </tr>
<%i++; }%>
<div class="abstractWriteup"><textarea name="abstract"></textarea>
<input type="submit" value="Add Spotlight"></div>
</table>
</form>
Thanks in Advance
Dean
You can easily move DOM nodes around using Node#insertBefore. (That link is to MDC, but it's a standard method and well-supported.)
Here's an example using Prototype, but you can do it with jQuery or other libraries, or just straight DOM methods like the one linked above (it's just more hassle without a library):
// Called at some point during page init to hook up the event handler
function hookRadioButtons() {
var form;
form = $('theForm'); // Assuming you put an ID on the form
form.observe('click', radioButtonClick);
}
// Event handler for radio button clicks
function radioButtonClick(event) {
var btn, div;
// Get the (extended) DOM element for the button
btn = event.findElement('input[type=radio]');
if (btn) {
// Get the (extended) DOM element for the div
div = $('theDiv'); // Assuming you gave the div an ID
// Starting from the button, go `up` to the table cell,
// then over to the following cell, and then insert the
// div at the top of it
btn.up('td').next().insert({
top: div
});
// Show the div (if it's hidden)
div.show();
}
}
That's completely off the cuff and untested (and full of assumptions), it's just to give you an idea.
you can use after(html) or before(html) you only need to add to every radiobutton a unique value but you already did that with value="<%=ids.get(i)%>"
something like this?
$("input[value='PutIdHere']").parent().after(htmlContent);
Move the div "abstractWriteup" to after the closing table tag. It's invalid code to have it inside a table without a containing row and cell anyway, and this will ensure that it appears below the table.
EDIT
I realise I may have misunderstood - you want the div to appear below the radio button which has been clicked, don't you?
You can't put <div> inside a table. It's invalid and browsers may render it in odd ways, like, as you said, as the top.
You would have to either:
close the table, put the div and then if you need more rows after that open a second table. To ensure two tables have the same row widths, you can set table-layout: fixed on the tables and add <col> elements with explicit widths for the non-liquid columns. Or,
include the textarea/submit in a new <tr><td>.
Having said that, I don't know why you're using a table here when you aren't really doing anything with the columns. A simple <div> for each radio would be easier to handle.
Since you will also only ever have one radio ticked, you can re-use the same textarea/submit elements, and just move them to the right row onclick. With JavaScript disabled they'll just stay at the bottom.
<form id="publications" method="post" action="...">
<% for(int i= 0; i<ids.size(); i++) { %>
<div>
<input type="radio" name="publicationIDs" value="<%= ids.get(i) %>" id="pid_<%= ids.get(i) %>">
<label for="pid_<%= ids.get(i) %>">
<%= info.get(i).replaceAll("&", "&").replaceAll("<", "<").replaceAll("\"", """) %>
</label>
</div>
<% } %>
<div id="abstractsubmit">
<textarea name="abstract"></textarea>
<input type="submit" value="Add Spotlight">
</div>
</form>
Note: <label> added, for better usability (can click on name to select radio). replaceAll here is acting as a poor man's HTML encoder. You need an HTML encoder every time you output text into HTML, otherwise characters like < in the text will mess up the page, and potentially create cross-site-scripting security holes.
Unfortunately, in an appalling oversight, the original version of JSP didn't include an HTML encoder. Many, many libraries for web-related software have one you can use to avoid having to write out all these replaceAlls. Also, if you're using JSP 2.0+, you should consider using JSTL c: tags rather than scriptlets. <c:out> gives you an output-with-HTML-encoding behaviour by default.
Example script to move the abstract element about:
<script type="text/javascript">
var radios= document.getElementById('publications').elements.publicationIDs;
function update() {
// Get div containing selected radio
//
var selected= null;
for (var i= radios.length; i-->0;)
if (radios[i].checked)
selected= radios[i].parentNode;
// Move abstract div just after it
//
var a= document.getElementById('abstractsubmit');
if (selected===null) {
a.style.display= 'none';
} else {
a.style.display= 'block';
selected.parentNode.insertBefore(a, selected.nextSibling);
}
}
// Bind to all radios, and reflect initial form selectedness state
//
for (var i= radios.length; i-->0;)
radios[i].onclick= function() { setTimeout(update, 0); };
update();
</script>
setTimeout is used here because the onclick event fires before the radio have been selected to reflect the click, so update can't yet read which radio is chosen. The right thing to do would be to use onchange instead, but unfortunately that doesn't fire at a useful time in IE, so onclick is what we're left with.
update is run at the start because when you're doing forms you can never be sure what your initial state is. Browsers often try to remember form fields when returning to a page, which may result in a radio unexpectedly being ticked at start of play.
(You could make this a little shorter with jQuery. But not massively shorter.)