Thomas Fuchs
Hi, I'm Thomas Fuchs. I'm the author of Zepto.js, of, and I'm a Ruby on Rails core alumnus. With Amy Hoy I'm building cheerful software, like Noko Time Tracking and Every Time Zone and write books like
   Want me to speak at your conference? Contact me!

In-Place Editing: The Summer 2007 Rewrite

July 17th, 2007

Christophe rewrote In-Place Editing from scratch, seeking to eliminate all the corner-case quirks, smoothe the API, clean up the options, and add several much-requested features. The result is now part of (as of changeset 7191)!

Does this break anything?

Well, mostly no. The only feature we know about that changed in a non-compatible manner is multiple-line management in the edited contents. Aside from that, existing options and behaviors remain.

Feedback welcome! I’d encourage trying out trunk—if you have any bugs or general feedback, please don’t hesitate to discuss this on the mailing list!


Many, many users monkey-patched the classes (that is, hacked internal methods directly, such as createEditField or createForm). This was never a safe practice, as it relies on undocumented, generally internal, and essentially unsupported aspects of the code. Note that you can of course continue to use the old version if anything blows, but I’d recommend to port the code over.

Because the code was heavily refactored in this rewrite, it is very much possible that some methods you used to monkey-patch are not there anymore, or that they still exist but do not have the exact same set of responsibilities as their previous version, or occur at the exact same moment in the in-place editor’s lifecycle. So you should beware, and when installing the latest, test your tweaks like crazy.

What’s changed

Aside from plenty of internal cleanup and performance boosts, there are essentially two changes to existing behavior: multiple-line management and a few options that morphed to something else. The sequence of AJAX calls used by such features as alternate text loading and final persistence have also been cleaned up, along with AJAX-related form disabling/enabling.

Multiple-line management

In the previous version, if the markup being edited contained any notion of multiple-line once rendered (e.g. it featured a <br/> or a <p>), you’d get a multi-line editor (a textarea). This behavior could be disabled by setting the handleLineBreaks option to false. The only way to get a single-line editor was to not feature such markup in the edited text (or disable its detection) and have the rows option set to 1 (one). The markup itself would be converted on-the-fly to multiple lines with no markup before being edited, but would not get converted back when being saved. Also, if the editor was configured to load alternate text from the server side, this conversion and detection would not occur on the loaded text, but on the loading… message cough.

We discussed this in length, and decided to change that a bit for better consistency. There’s no more implicit, one-way HTML conversion or what-not. If you want a multiple-line editor, just set the rows option to a value greater than 1 (as before), or just feature line breaks (actual line breaks, not specific markup) in the element’s contents. That’s it. If you want to make that content editable using an alternate syntax, use the alternative text options (loadTextURL and loadingText), which serve exactly that purpose.

There’s also a new option, called autoRows, which determines how many lines are used when rows uses its default value of 1 (one), but the element’s contents features line breaks. This option defaults to 3 (three).

Option changes you should migrate to

Several options changed their names and value sets with this rewrite. Here’s a rundown:

  • okLink and okButton became okControl. The accepted values are 'button' (default), 'link' and false.
  • The same goes for cancelLink and cancelButton towards cancelControl, except the default value is 'link' here.
  • highlightcolor and highlightendcolor were re-cased to highlightColor and highlightEndColor.
  • evalScripts was a source of confusion, was improperly used internally, and has been replaced by htmlResponse (see below, “New options”).

Yes, you can’t have a button and a link for the same action anymore. But then, nobody used that configuration (and frankly, who would?!).

Currently, a deprecation layer lets you still use the former option names and values, but that won’t be kept in eternally, so migrate your scripts now!

What’s new

Plenty of stuff!

New options

  • As stated earlier, okControl and cancelControl govern OK/cancel controls in the form now.
  • fieldPostCreation determines what to do after the editing form gets ready. It can be 'activate' (default), 'focus' or false.
  • htmlResponse determines what kind of request to perform when the editing form gets submitted to the server:
    • If set to true (default), an Ajax.Updater is used internally, with its evalScripts option defaulting to true (you can change that using the ajaxOptions option). Only successful requests will update the element: failures will trigger the onFailure callback.
    • If set to false, an Ajax.Request will be used, as we assume the server will not return an updated version of the contents for our element: we’re just notifying it. However, as always with Ajax.Request, defaulting to get method (once again, ajaxOptions can change that), returning a JavaScript-typed response will trigger automatic evaluation. Failed requests trigger the onFailure callback, as always.
      This essentially replaces the former evalScripts option, whose named introduced confusion.
  • rows is now complemented by autoRows, as mentioned earlier: if rows is 1 and the value has line breaks, a multiple-line editor with autoRows rows (defaults to 3) will be used.
  • stripLoadedTextTags determines whether to strip the tags from the alternate, AJAX-loaded contents (defaults to false).

 New/changed callbacks

In-place editors feature a variety of callbacks, some of which existed in the previous version. Their exact triggering moment in the lifecycle may have changed, though. Here’s a full rundown so there’s no doubt left.

  • callback is called when the form is about to be submitted to the server. At this point in time, the form was taken out of the DOM already, and the saving text was displayed. This takes the form and the editor’s value as its arguments, and defaults to serializing the form using Form.serialize (which makes it easier for you to customize the form’s contents).
  • onEnterEditMode triggers right before in-place editing begins, basically as soon as the editable element is clicked. The external control, if any, is not hidden yet, nor is the element, and the editing form hasn’t yet been created. It takes the in-place editor object as its single argument. Along with a few others, this was previously not a proper callback (passed through the options hash), but a method you had to override in your own version of the class. Ugh.
  • onLeaveEditMode is symmetrical, and triggers once the whole editing process is over and every part of it is out of the page’s DOM. Same argument as onEnterEditMode.
  • onEnterHover triggers when the mouse enter the surface of the editable element. This takes the in-place editor object as its single argument, and defaults to setting the background’s color to the highlightColor option.
  • onLeaveHover is symmetrical, triggering when the mouse leaves the element’s surface, and fading the background color back to its original value (keeping the background image, too).
  • onComplete triggers when the editor form is done with, either through cancellation or successful submission to the server, and reacts by performing a highlight (keeping the background image). It takes two arguments: the XMLHttpRequest object (or undefined if it’s a cancellation) and the editable element.
  • onFailure triggers whenever an internal AJAX request fails (including HTTP responses with non-2xx codes). It takes the XMLHttpRequest and the in-place editor object as arguments, and defaults to displaying a message box with the details of the response’s body.
  • onFormCustomization is a brand-new callback that aims to make it easier to customize the editor form. It is triggered right after the editor field was created, and before any controls (OK/Cancel) are in. As it gets passed the in-place editor object and the form as its arguments, it’s pretty simple to use appendChild in order to add a few custom controls in there.

Support for multiple in-place editors

Every single AJAX request made by an in-place editor guarantees a new editorId parameter, that contains the edited element’s id attribute. This lets you factor the code for multiple in-place editors in a single server-side resource.

A better collection editor

The previous version of Ajax.InPlaceCollectionEditor required that you provide the collection of items as an array in the collection parameter at construction time. What’s more, it did not integrate well at all with alternate text loading.

The new version plays nice with alternate text loading, and also lets you dynamically load the collection through AJAX, every time the editing form is created (as always, the collection query will feature a editorId parameter, as described earlier). The two allowed formats for the collection are retained: either an array of values, which will serve as both value and text, or an array of two-item arrays, the first one being the value, the second one being the text.

When loading an alternate text (which will be used to match against item values in the collection), the loading text is featured as the sole option in the temporarily disabled dropdown list. The same kind of display is used while the collection is dynamically loaded. Note that the collection is loaded before the alternate text is.

The value option, which could force a given item in the dropdown list to be selected regardless of the current contents of the editable element, is still featured (however seldom it is used).

So, start your In-Place Editor engines!