Freemarker and Struts 2 error handling integration - struts2

I have a project where Struts2 provides the MVC layer and FreeMarker templates are used update the web pages. Some of the FreeMarker templates returns partial views - for example addNewRow.ftl which returns a table row.
Actions are validated and error messages are added to web page using addActionError method.
In the cases where the web page only receives a partial view, the validation messages aren't getting displayed since the error info is missing from the page.
How can you retrieve and show validation errors occurred in the operation in this case?

Just beacuse you elect to redraw a portion of your UI based on a user operation doesn't not mean that is where all that information must be presented. With the use of jQuery, you could easily take error information and insert it into any existing DOM node on the page, even those that aren't part of the portion of the page you just rendered.
Another solution is to redraw a larger portion of your UI instead that would include the error display container and simply serialize the action messages out during the redraw.
Or you could look at your redraw callback and see whether it can be split into multiple invocations that could each handle their portion of the page redraw independently, e.g: add your table row and render error messages.

Related

How does rails handle form validations?

When a rails form fails on the the front end, how is the javascript handled? e.g. a numericality error fails or a required: true option is not fulfilled.
I would like to add a callback function to this. Is there any way to execute some javascript on the failure?
There is no javascript involved on the rendering of the errors on a rails form. The way the errors are displayed are determined by a function called field_error_proc, specifically ActionView::Base.field_error_proc. You can check Ryan Bates' railcast about it
If you are submitting your form via AJAX, you can attach a listener for the ajax:success event. Otherwise the only (ugly) way I can imagine is to do something like:
$(document).on('app:form-has-errors', function() { //do what you have to do} and on the view
<script>$(document).trigger('app:form-has-errors');</script> if the object in the form has errors (#your_object.errors.any?).
I don't think it is possible in another way since the submit and re-render on errors are complete http requests that will clear any events you might bind to the document.
Rails built in validation doesn't operate on the front end (client-side), it is done on the server.
If you want to do javascript validation you will need to find a gem for this or use a jQuery library, or write the validation code yourself.
The normal way form validation is done when using vanilla Rails is like this
The invalid form content is submitted to the controller
The controller uses the content to make the ActiveRecord object
The controller tries to save the object to the Database, kicking off a set of steps that starts with checking the object's validity
The model's validators say it's not valid and "decorate" the object with error messages specifying in which ways it is not valid.
The controller sees that the save failed and renders the view that had the original form in it, providing that view with the object, now with error messages.
The view when it renders the form checks the object for error messages, if it sees any it helpfully displays them to the user. It also populates the fields with the data that the user had previously entered that it gets from the object.

making unobtrusive validation work when using Select2 ASP.NET MVC

Select boxes converted to Select2, do not automatically integrate with unobtrusive validation mechanism in ASP.NET MVC framework.
For example, on a form which contains a regular select box (marked as required in model definition), submitting the form while no options have been selected in the select box, will cause the border and background of the select box to take a reddish color, and by using #Html.ValidationMessageFor, error messages, if any, can be displayed beside the box. However if the select box is converted to a Select2 component, then none of the mentioned features work any more. Even the validation error message will not show up.
It seems that the reason for even the validation error message not showing, is because Select2 changes the display CSS property of the original select box to none (display:none), and I guess the unobtrusive validation script does not bother generating error messages for invisible fields.
Any ideas / solutions?
This issue isn't really specific to Select2, but rather to the jQuery unobtrusive validator.
You can turn on validation for hidden fields as highlighted in this answer.
$.validator.setDefaults({
ignore: ''
});
As the comments noted, it didn't work inside an anonymous callback function within $(document).ready(). I had to put it at the top level.
I've run into similar issues with the select2 plugin. I don't know exactly which features you're using specifically, but in my experience, when you set an element as a select2 in the document.ready event, the plugin will change some of the element's attributes on the fly (inspect one of the elements after your page has finished loading - oftentimes you'll see the id and class properties are different than what you're seeing when you view source).
It's difficult to offer more without actually seeing the code, but here's a few ideas to get you started:
First off, obviously make sure you have the a link to your select2.css stylesheet in the header.
Then, since you're talking about form submissions, I'd recommend you examine whether or not you're getting a full postback or submitting via AJAX (if you're using jQueryMobile, you're using AJAX unless you override it in the jquerymobile.js file or set a data-ajax="false" in your form attributes). You can just look at the value returned by Request.IsAjaxRequest() for this. Obviously if you're submitting via ajax, you won't hit the document.ready event and the select2 won't initialize properly and you'd need to figure out a way around that. Try refreshing the page after the submit and see if it renders the select2 component.
Then I'd suggest examining the elements and see if they're not behaving like you'd expect because you're actually trying to work with classes that the plugin has reassigned at runtime. You can either just adjust your logic, or you can dig into the select2 code itself and change the behavior - it's actually fairly well-documented what the code is doing, and if you hop on the Google group for select2, Igor is usually pretty quick to follow up with questions.
like this
$('select').on('select2:select', function (evt){
$(this).blur();
});
$('body').on('change', 'select.m-select2', function () {
$(this).blur();
})

Is JSF 2.0 View Scope back-button safe?

Is the JSF 2.0 View Scope "back button" safe? e.g. if I store a model in View Scope and go from page 1, page 2, page 3, to page 4, modifying the model object along the way (via input fields), and then hit the back button twice to go back to page 2 and make changes (taking me again to page 3), will the model in view scope have only changes that were made when page 2 was originally rendered or will it have later pages' changes?
Oracle ADF had/has something called "process scope" that handles this by tokenizing what is placed into session, so each page has its own copy of the model.
To start, the view scope is bound to a particular page/view. Multiple views won't share the same view scoped bean. The view scope starts with an initial GET request and stops when a POST action navigates with a non-null return value.
There are in general the following scenarios, depending on whether the browser is instructed to cache the page or not and the JSF state saving configuration. I'll assume that the navigation between those pages took place by a POST request (as it sounds much like the "Wizard" scenario).
When the back button is pressed:
If browser is instructed to save the page in cache, then browser will load the page from the cache. All previously entered input values will reappear from the browser cache (thus not from the view scoped bean in the server side!). The behavior when you perform a POST request on this page depends further on the javax.faces.STATE_SAVING_METHOD configuration setting:
If set to server (default), then a ViewExpiredException will occur, because the view state is trashed at the server side right after POST navigation from one to other page.
If set to client, then it will just work, because the entire view state is contained in a hidden input field of the form.
Or, if browser is instructed to not save the page in cache, then browser will display a browser-default "Page expired" error page. Only when the POST-redirect-GET pattern was applied for navigation, then the browser will send a brand new GET request on the same URL as the redirect URL. All previously entered input values will by default get cleared out (because the view scoped bean is recreated), but if the browser has "autocomplete" turned on (configureable at browser level), then it will possibly autofill the inputs. This is disableable by adding autocomplete="off" attribute to the input components. When you perform a POST request on this page, it will just work regardless of the JSF state saving method.
It's easier to perform the "Wizard" scenario on a single view which contains conditionally rendered steps and offer a back button on the wizard section itself.
See also:
javax.faces.application.ViewExpiredException: View could not be restored
What scope to use in JSF 2.0 for Wizard pattern?

ASP NET MVC: Dynamically adding or removing inputs on the form - unobtrusive validation

Before starting, I do have a very particular question and if you want to answer it go straight to the end. But I do welcome comments and advices hence the lengthy post.
OK, we deal with a lot of forms and some of these forms are quite lengthy and have many fields. We also have a requirement - in addition to top level fields - to be able to have variable number of repating rows - as we call them. For example, let's think of a customer which has name, surname and age while it can have zero or many addresses (say 0 to 10) so the user must be able to add or remove contacts from the form while filling it in. So typically user gets and "Add" button to add more addresses and next to each address, a delete button. Potentially there could be more than one repeating section in the same form but I am not going there. The point is, because of legal and historical reasons, all the forms must be saved at once so while the forms can be edited, we cannot accept a half-filled form and have another page for users to add and remove addresses, e.g.
I am using ASP NET MVC 2 (strongly typed views with a single generic controller) with client side validation and heavy jquery scripting for flashy features. We are probably going to migrate to ASP NET MVC 3 very soon and I am already playing with 3 for finding a good solution. These addresses are defined on the Model as List<Address>, e.g.
I currently have a working solution for this issue but I am not satisfied with it: I have an HTML Helper that names the add or delete buttons and a bit of JavaScript to disable validation and allow the form to be posted back (even invalid) and since I can find out the name of the button that was clicked, I have all the necessary logic to handle add or delete and works really well.
But I am posting back and the form is reloaded and I am looking for an aletrnative solution. Here are what I can do:
Do everything in the client side. "Add" button will clone one of such addresses and "Delete" button will remove() the element. I only have to rename the indexes which I have done. We were using jquery calendar and it was breaking on the new elements which I have also fixed. But the validation is not working which can probably work with ASP NET MVC but this solution looks like a very brittle one - a house of card which looks great before you add another card.
Post the whole page usin Ajax and then load it back again: This is probably better than my current solution but only slightly.
Use ajax to post the form and get back JSON and use the data to build the elements or remove them: Again a house of card because of extensive client side scripting
Serialize the form and post using Ajax to a particular action and get back only the repating section (as a partial view). The action on the controller can be reused and called from the view itself to return the partial view
OK last one is the one I am working on but there is an issue. ASP NET MVC 3 with unobtrusive validation works only if the form is engulfed in a BeginForm() while my top level view has a BeginForm() but not my partial view. It works well when I call it from the view but not on the ajax call to get just the repeating section.
(Question)
So is there a way to tell ASP NET MVC 3 to spit out validation data atttributes regardless being in a BeginForm() block?? To be honest if this is not a bug, this is definitely an important feature request. I have in fact used reflector to disassemble the code and the condition seems to be there.
Short Answer:
Add this to the partial view:
if (ViewContext.FormContext == null)
{
ViewContext.FormContext = new FormContext();
}
I don't think it is possible using the default unobtrusive libraries supplied. If you look at jquery.validate.js and jquery.validate.unobtrusive.js it looks like it only validates what is inside the form.
There's a few posts about it if Googled and a few work arounds.
I had a similar issue (although much simpler) where I had a validation summary at the top of the page and multiple forms but the unobtrusive javascript would only populate the view summary if its inside the form (jquery.validate.unobtrusive.js line 39 if interested...).
I'm not sure if the validation library is extendible but most things in jquery are so that might be an option if you want to go down that road.
As far a possible solution to your problem I'll put in my 2 cents for whats its worth.
You could have two actions that are posted to. The first action is you post your model with no js validation and all validation is handled in the code - this will catch all user with javascript turned off.
Your second action is you serialized the model. In mvc 3 using the Ajax.BeginForm has an AjaxOption for Url where you can specify an action for the jquery to call (where it serializes the form form you and you can decorate your action with your strongly typed model). Here you can check the model and return a json result and handle this in the javascript.

How does validation in ASP.NET MVC 2 actually work?

I am trying to trace through why my ASP.NET MVC 2 validation isn't working, but I cant find enough about HOW it works to be able to do this.
I have followed the steps in this useful article by David Hayden which seems to be the best documentation currently out there, but nothing actually happens.
I get validation when i submit to the server (as I did since Preview 1 when i added data annotations to my model) but I'm not getting any client side validation.
How can i trace through to test? So far I have verified the following obvious things
MicrosoftMvcJQueryValidation.js and jquery.validate.min.js files are being downloaded
Html.ClientValidationEnabled = true
I cant see easily what is hooking up to which events to know quite how to debug it.
Here's what I've learnt:
MOST IMPORTANT
Your HTML Form must be created with the using directive, not just BeginForm and EndForm.
You must set Html.ClientValidationEnabled = true BEFORE you start your 'Form'
You must use Html.ValidationMessage for each field
You must set Html.ClientValidationEnabled = true on each partial control (ascx)
HOW IT WORKS (very simple overview)
When you do Html.BeginForm it creates a 'FormContext' in the ViewContext
When ValidationMessage helpers are used, metadata is put into the form context
When the form is disposed (by the using statement) it writes out all the validation code
MISC
I cannot seem to get validation working when I have a partial control, if that control uses a different model from the view that defines the Form.
You do NOT need to use Html.TextBoxFor or Html.ValidationMessageFor, you can use Html.TextBox and Html.ValidationMessage
In order for a field to be validated client-side you have to specify a call to Html.ValidationMessage/Html.ValidationMessageFor<> for the field (just like David did in the tutorial you linked) within the view. This is essentially a trigger to the client-side validation logic that you want to run validation for that field.
If there are situations where you don't actually want a validation message to visually appear for each field (i.e. by using Html.ValidationMessage), but would rather allow a summary to be the sole source of validation error messages (i.e. by using Html.ValidationSummary), you still need some way to "trigger" the validation to occur for the specific fields you want it to. This can be achieved by using the Html.Validate/Html.ValidateFor<> methods within your view. Those helpers won't render anything, but will simply register the specified field for client-side validation.
Both of those requirements exist since you might not want the client-side validation to automatically validate every property on your model object, since some of them might not even be part of the form that you're wanting validated.

Resources