This will indicate when a password does not meet the validation criteria defined in the model. You can add custom validations to check the strength of the password itself.

I don’t really like this approach because I don’t want to clutter up the session with various form parameters. Thankfully, there is a simple way to do this by once again abusing the built in text_field_with_auto_complete function…

[In my case, I used the :extra data to exclude certain objects from the auto_complete_for response]

By passing the :url option, you rewrite the URL that the AJAX actions call, enabling you to add extra parameters that can be acted on by the method. Also note that you can change the name of the action called (in fact, you need to specify it or it will default to the name of the action for the current view… probably not what you want). This is handy because it lets you assign the response to a variable with a different name without writing a separate auto_complete_for handler. In this case, the submitted form will have a ‘new_object[name]‘ in the params, even though the action called was ‘auto_complete_for_object_name’.

Needless to say, you need to write a custom ‘auto_complete_for’ method to handle the extra data. Note that you could also use this method to handle multiple auto_complete fields.

The standard rails approach to handling form validation errors is to spit out a long list of all validation errors in a box at the top of the form. The problem with this approach is that there is a separation between the list of errors and the field that spawned them, like this.

This can be a bit annoying, particularly in a long form where the error message and the offending form field may not appear on the screen at the same time. Notice that you have to first read the error message, and then read the titles of all the fields to find the offending one(s) that need to be corrected. They aren’t even in the same order as they were specified in the form, which makes it even more confusing. This is just extra work for the user, particularly if your form is long or complex.

One way to address this problem is to put the appropriate error message right next to the field with the error. This way, it’s pretty obvious where the problems are, and even what they are. You can also generally avoid using the name of the field in the message.
A simple example of this might look like this….

This still isn’t pretty, but at least the user knows exactly which fields need correcting, and a little CSS magic can even place the message to the right of the offending field.

So how do we implement this?

Simple… first go into your view and delete the

line.

Then stick this in your application helper.

Then in your form view, add an ‘error_for’ call wherever you need one…

If the model fails a validation test, then it will show the message right next to the field that caused the validation problem.
Also note that if you define an instance variable called @errors containing a hash of field_names and messages, they will also be used. This is handy for those form fields that don’t correspond to a model attribute.

Stay tuned for Part 2, where I will describe a how to make these error message dynamic with AJAX.

This technique is clearly one to keep an eye on for a good user experience. It is not yet clear to me when this type of interface can be used for maximal benefit, and when conventional paging should be the preferred method.

Unspace – Endless Pageless: No More Next Page

Yes, that is really me.

Occasionally there is a need to have a user enter a unique string. For example, if you want to ensure that a new user’s login id is not used by someone else. The traditional rails way of doing this is to set up a ‘validates_uniqueness_of’ validation on the model and then catch the error when you try to save the record.

From the user’s point of view, this can be a horrendously painful experience that can be akin to trying to outsmart the Sphinx.

The biggest problem here is that there is no feedback to the user if the value they selected is valid until they submit. Some forms will also force you to retype your new password everytime this fails as well. Double ouch!

Solutions

1. offer suggestions for valid responses when it fails the first time. Not exactly the best, you might as well autogenerate their id for them and hope they don’t forget it.
2. provide feedback with AJAX to let them know if the value is ok.

As you might have guessed from the title of the article, I’m going to show you how to do this with a standard rails auto_complete text field.

That’s it. Anytime the auto_complete matches a record, it will change the ‘login_error’ div to indicate there’s a problem.

Net result…. a much better user interface experience and much happier users.

Then just use ‘commit_tag’ instead of ’submit_tag’ in your form.
Strictly speaking, you don’t really need to give the button an ‘id’, but I do because it has the bonus side effect of making it easy programmatically disable the button from RJS templates.

One reason this approach is superior (I think) to an AJAX call is that there is less of a chance that an over-enthusiastic user will be able to click the button again before the AJAX response renders.

This requires some sort of javascript hackery to achieve, but the easiest way I’ve found to do this is with my trusty old KRJS pocketknife.

This will add this behavior for all submit buttons with the ‘commit’ id. The hardest part here is remembering to add the ‘commit’ id.

Requires: KRJS Plugin

UPDATE: just to clarify this a bit. If you have more than one auto_complete that tries to access the same model/field combo, you get into trouble. You can have multiple auto_completes so long as they access different model/field combinations.

I tried a couple of hacks around this problem (some involving method_missing), and finally found a method that works pretty well and doesn’t mess up your controllers.

The approach I ended up using was to rewrite the ‘auto_complete_for’ and ‘text_field_with_auto_complete’ functions so that they ignore trailing numbers on a dom_id. Once you install the plugin (see below), you simply use it like this…

All the auto_complete fields named /object[-_]d+/ will call the ‘auto_complete_for_object_name’ function in the controller now. The default response function will pull out the correct information and respond as usual.

When you submit the form, the values will be passed in the params hash as

Installation

If you have a ‘Really Long List’ to select from, then using a cascading select makes a lot of sense (unless you want your users to hate you). The problem is that they are a pain to code (well, not so bad if you use the KRJS plugin ), and they are still a pain to use if you actually know what you are looking for (then you have to remember what ‘category’ if falls in).

Lately I have been getting around this little problem using auto-complete boxes.
In Rails you can have an auto_complete box act like a select like this…

Now when the user starts typing in the auto_complete field, they will get records that match against both the name and the category of the record. The enterprising reader could also replace certain characters (like space) with the SQL wildcard ‘%’ to get more flexible search results.

A user might be able to just type: ‘f mus’ to find ‘Ford – Mustang’.
(It’s not well known, but ‘f%mus’ will match ‘Ford – Mustang’ using the standard auto_complete box that ships with rails).

This type of entry is really fast and quite intuitive once users get the hang of it.

I find that this system works best when
1. There are a ton of choices
2. The user already has a pretty good idea which choice they want to select, but don’t really want to spend 15 minutes scrolling to find it.

It’s not particularly good if the user has no idea what to enter. In those cases, the standard cascading select may help walk them through the selection process.

Notes: You can avoid using the ‘full_name’ column if you properly design your own custom auto_complete_for function