HTML Message Ajax Design Pattern

Having more and more browsers supporting XMLHTTPRequest object to send the data to server and get the response from server without reloading the page, it is possible to refresh only a part of the page. There are various techniques to do this. One such technique is that the server send the XML back to the browser and the Javascript on the browser will convert the XML into HTML by creating the DOM nodes on the fly. This is useful but has some limitations.

* If the Javascript is turned off, there is no graceful degradation of application
* Since the Javascript is running on the browser (client side), changing the structure of the generated HTML is hard
* This is not feasible when HTML generated is complex

One can overcome these limitations by sending the HTML snippet in the response. Instead of sending the XML back to browser, generate the HTML on the server side itself. On the browser side, replace the content of the DOM node with the response HTML response from the server.

If the Javascript is turned off on the browser, user will see at least the content of the HTML on the browser instead of just the XML content. So there is a graceful degradation of service on the client side

If HTML generated need to be changed, application is not dependent on changing the Javascript and hoping that the client will reload the page and will get new Javascript. Even if the Javascript is changed, some web proxies may not pull the new Javascript, resulting in the browsers (clients) behind the proxies will continue to get the old Javascript and old HTML will get rendered.

There are various frameworks like Symfony, Django make it possible to generate the complex HTML possible with least effort. In case the request is not made via XMLHTTPRequest, it is possible to send the whole page (with header, footer etc.) in the response. If the request is made via XMLHTTPRequest object, send the response without the layout decoration (just the HTML snippet). Javascript running on client side will replace the content of the DOM node as required.

Take for example, there is an application where the page loaded has many graphics and only a part of the page changes when user take some action like filling a form for sending the email. It is absolutely unnecessary to generate the whole page again just for a simple action of sending an email. I will make an attempt to explain with the example of sending the feedback via a form embedded in the page. I will try to explain this with PHP Symfony framework and Javascript (AJAX).

The URL for the page is /new-page. This page has a feedback form in it. The action of this form is handled by /feedback. Server side code for /feedback expect a form having a text content, sends an email to the product manager and send same page as new-page but replacing the form with thank you note.

Below is a code for the new-page action.


Code for new-page response.


Now is the magic of jQuery Javascript library to submit the content to /feedback url and use the response content to replace the HTML form.



Server side code for handling the post from the browser to /feedback



HTML response from /feedback


In Symfony framework it is possible disable/enable the layout during run-time via setLayout() method of sfAction instance. Setting the parameter to false will disable the layout and send only the content of the action template back to the caller.

There you go.. A simple way to improve the usability of the site even without Javascript support.

Execute action within another action

In the current project I am working on (using Symfony framework), I came across a situation where I had to call one action within another action and capture the content of 2nd action. This content is included within the result from 1st action. After doing some research found that it it possible to do so by calling getPresentationFor method of the controller object.

Within the action's execute method,

$content = $this->getController()->getPresentationFor($module, 'my2ndAction');
$this->content = $content;

File System Over SSH

Some of the development I am doing require me to do the development on some remote machine. Since the development is done in Java, I need Eclipse environment. On the remote system I cannot run Eclipse as it is a server box and does not have display attached to it. I found an elegant solution for this. Thanx for Macfusion

I can mount remote machine's directory on my local machine and work as if it is a local directory. All the dirty work of communication is taken care. I don't need nfs mounting which is not secure. Macfusion let me mount the directory on remote machine over ssh

Why Erlang scares managers?

Managing is all about control. Managing is about knowing (almost) everything about what one is controlling. If the team is using something manager don't understand or willing to understand, manager is loosing control. Again it is all about control.

99% of managers are comfortable with technologies like PHP, Java etc. Languages with new ideas like Erlang, LISP are far fetched for many managers. They don't want to learn about these new technologies or new ideas. Having these new technologies in the team force them to come out of the comfort zone and take the control away from them. This is not a good news for them. One argument I have heard every time I make a case for why Erlang (is better in-terms of hardware utilization and lines of code) is that "It is hard to hire Erlang developers than PHP/Java developers'. As I see competent developers want to learn new things than putting themselves in the same-old-same-old world. Competent developers are capable of learning new technologies and new ideas.

Off course, there is a initial learning curve. It is always there anywhere you go. Developers/Engineers need to learn the new way of doing things anywhere they go. Some companies are willing to give more time to developers to get familiar with the internal technologies and some don't. But it is a absolute necessity. Everyone does not follow the same development methods or procedures. In a big organization, this changes from group to group.

To stay or to leave....

There comes a time when one has to stop doing the things one was doing for along time. It is hard to do that... Sometime it feels it is impossible.. Some people lack courage to do that, some people lack motivation and for some people situation don't let them..

Once the decision is taken, is it possible to revert that decision.. Yes.. I think it is possible to that. Question is what it the cost of reverting the decision. I don't think it is much..

Dog is not Cat

While talking to my friend today, something struck me regarding the notion of object oriented programming, hierarchy of classes and how these concepts are over used.

Objected oriented design is heavily used for thousands of years. For example, constructing a home. If you think about it, home is built in a very modular way (aka object oriented way). Every part of home is a object [door, window, wall, roof etc.]. They can be put together in many ways to construct a home. The final shape of home may change, but the the basic building blocks remain the same.

Now to the point of using this in computer programming. I think object oriented programming is misunderstood in the software industry. People seems to confuse reusability of the code with object orientation. For example, take the hierarchy of animals. Animal is a base class, cat is derived from class Animal. Now let's say I want to declare a new class Dog. Since Dog class was not considered while implementing Cat class, common functionality of Animal (like see, listen etc.) are implemented in Cat class. Now when one want to design Dog class, since Cat already has the functionalities required by Dog, derive Dog from Cat. This is so wrong. Logically when you derive one class from another, you are implying "is-a" relation. That means, by deriving Dog from Cat [just because you have functionality required by Dog is implemented in Cat], you are implying that Dog is a Cat.

Cases like these, it is better to sit back and think a little. May be code re-factoring will help here. One can move the functionalities that are required by both Dog and Cat [may be Elephant, Tiger etc.] can be moved into the Animal class and implement those functionalities in terms of some abstract concepts like eyes, legs, ears etc.

Now enough of messing your brain.. Go back to work and start thinking about the designs you have done earlier and how you could have done it better [not that what you did earlier was wrong].

Hacking search suggestions

Opensearch specification has extension for search suggestions. Major search providers have their own search suggestion entry points. After looking doing a search of my own, I found the entry points for the 3 major search providers

Bing - http://api.search.live.com/osjson.aspx?query={Search Term}
Google - http://suggestqueries.google.com/complete/search?q={Search Term}&client=firefox
Yahoo - http://ff.search.yahoo.com/gossip?output=fxjson&command={searchTerms}

For Google entry point, removing client parameter will also provide the number of results in the response. This format is not as per the Opensearch standards.