I know, it’s been too long. What can I say? I’ve been busy. Really, really busy.

A few weeks ago Cloudera announced our latest release of our Hadoop Distribution and the open sourcing of numerous projects that are part of that distribution. If you’re reading my blog because you care about JavaScript and stuff, this won’t be terribly interesting to you (if you’re into distributed computing though, that’s another story). What will interest you is that one of the things we open sourced is Cloudera Desktop, which is now the Hadoop User Experience (aka “HUE”).

Why is this interesting to you? Well, HUE is the thing I’ve been working on for over a year now at Cloudera and I can tell you that the JavaScript foo inside it is some of the most interesting stuff I’ve ever written. Here’s a little preview:

• JFrame: Essentially a JavaScript implementation of an IFrame. You assign a url to a DIV and now all links, forms, etc. load in that DIV. Click a link, the div updates. It powers the Ajax windows on the Desktop UI and allows us to build plain old vanilla web applications and load them into windows in the browser and they all act like their own little browsers. Because they aren’t really IFrames, you don’t have all that history session clutter, you don’t need to load all your JS and CSS for every one of them, and they can easily communicate with each other, allowing applications to spawn other apps and continue talking to them.
• Behavior: Anyone remember Behavor.js from 2005? Well, this script is a bit of a nod in the direction of that early JavaScript tool. What it does is turn vanilla HTML with some declarative mark-up into interactive, slick looking UI goodies. For example, there’s a Behavior Filter for forms that make them submit via Ajax – the filter just finds any form tag with an HTML5 data attribute that looks like <form … data-filters=”Form.Request”>… and configures an instance of Form.Request for it. Same thing for sortable tables, input hint text, and other UI widgets.
• MooTools ART: HUE makes a liberal use of MooTools ART for windowing, nice looking buttons and inputs, icons and more. There’s still a lot of CSS stuff going on of course, but HUE looks as good as it does in large part due to ART. The code is pretty stable. You can download and play with the ART Widgets to see how it works. Before we can release all this stuff though we need to document it all and so it’ll likely be a while longer before there’s an official download on MooTools.net.

There are some more detailed posts about HUE on Cloudera’s blog, one that gives a nice overview of what it does and one that I posted today that gives an overview of how to develop your own applications on it.

It’s worth noting that HUE, as it is right now, is focused on supporting application development for the Hadoop ecosystem, but there’s nothing stopping it from being used for something else. The Desktop system itself doesn’t depend on Hadoop. If I had the time to make it stand alone I’d call it the MooTools Desktop. If you find yourself wishing you had such a thing, don’t hesitate to fork the code and work on it or to hop on the HUE Developer Google Group to ask questions or make suggestions.

Now that most of my work is open source I hope that in the coming weeks and months I’ll be able to post more often. When ART is released I’ll probably write about how to use it more. Right now the Behavior framework is part of the ART-Widgets repository but maybe I’ll release that sooner if enough of you are interested in it…

I know, it’s been a while. A long while. What have I been up to? Oh, a little bit of this and a little bit of that and maybe a pinch of this.

And speaking of what I’ve been up to, I’ve mostly been focused on my job at Cloudera, which is teh awesomes. How awesome? Well, I work in an awesome space with awesome people on awesome projects.

…and we want you. Were looking for someone to help us build a lot of cool things, and the job involves a lot of JavaScript. We push code into various MooTools projects daily. It’s all built on MooTools ART, ART Widgets, MooTools More, and of course Clientcide’s JS, as well as other such goodies. You’ll be hacking with the cutting edge of MooTools and pushing your work back into the community.

Here’s our sales pitch:

There’s more to UI development than writing a bunch of HTML, CSS, and JavaScript. You know what we mean; it takes design sensibilities to do it well. It’s the kind of thing you know when you see it – when things work smoothly and intuitively. Cloudera (recently voted as the #1 startup by VC Journal – http://cl.ly/fb5cc4338f0d07fc5539 ) is looking for someone with those kinds of sensibilities to help us build Cloudera Desktop – a virtual web-OS style application suite for managing large Hadoop data clusters. We want to make working with a virtual cloud of computers as slick as OSX or Ubuntu and to do it we’re writing some cutting edge JavaScript, but the people writing it need to have some design sense, too. If you enjoy spending time in Photoshop and Illustrator as much as you enjoy writing closures and recursive functions, then you’ve found the right job.

Required Skills:

• 3+ years of experience in developing web UI applications.
• Solid grasp of cross-browser, object-oriented JavaScript (closures, inheritance, etc)
• Significant experience in developing AJAX-enabled web applications
• Experience in coding standards-compliant semantic markup in XHTML/CSS
• Knowledge of and a preference for JavaScript frameworks such as MooTools, jQuery, YUI, Prototype, Dojo, etc.
• A passion for writing super optimized and clean code
• Some experience designing user experiences; a working knowledge of Photoshop and Illustrator

Strong Plus:

• Experience with data and information visualization is a strong plus
• Experience building web applications in Django/Python
• Experience developing desktop applications UI (e.g. with Adobe Air)

Involvement in an open source project (examples: committing code, writing plugins, involvement in user forums) preferably one focused on JavaScript

If you’re interested, ping me and send me your resume.

I’ve written a lot about MooTools over the last several years, the mootorial, my own book, not to mention all the posts here and loads of documentation. So you might think that I’d look at another MooTools book on the market as competition but you’d be wrong. I love that there are other people out there who want to help others learn MooTools. Writing tutorials and books is just as valuable to the framework’s growth as writing code. In the last year there have been a lot of new materials on that front now that I think about it. Ryan Florence has dedicated his blog to the topic. Mark Obcena has written a whole series called “Up the Moo Herd” that covers a lot of tips and tricks and best practices. Not to mention all the great work David Walsh continues to do (for instance, his recent post on NetTuts – “Make your MooTools Code Shorter, Faster, and Stronger” - was terrific).

So yeah, another MooTools Book? AWESOME. Read the rest of this entry »

At the beginning of 2009, MooTools was a single thing – one library maintained by maybe 5 or 6 people who actively pushed code into it. It’s hard to reconcile this with where it is now. Let’s see, first, MooTools was split into two in February – MooTools Core and MooTools More. The principal developer, Valerio Proietti (the creator of MooTools) focused most of his energy on MooTools Core with the help of several other developers, and I’ve focused most of my efforts on MooTools More, the official plugins for the library. While Valerio and others continue to refine the core, the -more team has brought the number of plugins for MooTools from around a dozen to nearly sixty.

Part of what changed was our active recruitment for people to be involved in our development community. Our developer mailing list went from about a dozen people on the mailing list with 4-6 people actively pushing code into MooTools to around 50 developers on our mailing list with somewhere around 15-20 people actively pushing code. These numbers tell part of the story; of tremendous growth – as much as 400-500%.

At the same time, there have been a lot of new things happening that may be slightly off your radar. For instance, a MooTools user named Piotr Zalewa showed up one day out of nowhere with a paste-bin app for MooTools that lets you create stand alone working code examples with HTML, CSS, and JavaScript which he launched at http://mooshell.net. Immediately upon joining the developer list he started adding features and working with others on the list to make it better and better and the result has been phenomenal. The usefulness of our user mailing list has gone up dramatically as users show up asking questions with working examples (or, more often, broken ones that they need help with). Other users can edit the shell and save a new version, responding with a link that shows exactly how to solve the problem. Daily we see how MooShell has affected our ability to help our users and we’re happy that MooShell is now an official MooTools project. Expect to see these working demos in our documentation early this year (to a demo of this in action just scroll down to the “Demo” section here).

There are a number of other new additions to the library that are still somewhere between pre-alpha up to beta states. There’s the Depender application that lets you build libraries on the fly. MooTools ART, a canvas adapter for MooTools, continues to come along. Slick, MooTools’ new selector engine is in testing phase and we hope to see it released early in 2010. We just released the MooTools 1.1 to 1.2 Upgrade Helper which should help those still using 1.1 to adopt the current release. MooTools 2.0 is still under heavy development, but we’ll be releasing MooTools 1.3 early this year with some new features. Even stuff that we don’t release, like a new search engine for mootools.net.

And then there’s the MooTools Forge, our new user-created plugin catalog. The Forge alone would be enough for me to call 2009 a big year for MooTools. It is absolutely what MooTools was really missing. Developer Guillermo Rauch cranked out a slick bit of technology that closely integrates with Github and makes it easy for developers to update and maintain their plugins. In only a month or so there already 100 plugins in there with more added daily.

All of these things, this proliferation of output and creativity, stem from one small change that happened early in the year. Something that in many ways no one outside our developer community would know about or even have noticed: MooTools went from being the work and property of one person to being a project owned by its community. For the roughly two years previous to 2009 that MooTools was around, it was developed almost entirely by Valerio. It’s true that other people, such as myself, contributed to the work with Valerio as the gatekeeper. The MooTools code has always been elegant and easy on the eyes to actually look at, and this grooming was a product of that single guardian who cared deeply that the code always be as clean and elegant as possible. It was uncommon for anyone to make a commit to the project without Valerio coming through and tidying it up, always making it just a little better and in the process teaching us how to be better developers.

In February of 2009, the first little step happened with the split of MooTools Core and MooTools More. The code in -more is still methodically groomed and maintained, but not by Valerio. The -more team does this service for each other, each checking over the others’ work and making it a little better, a little leaner. As we actively recruited people into the developer list, we had more and more people working on different things. Selector engines, paste-bin applications, plugin libraries, canvas adapters, a search engine for mootools.net, test frameworks, and more; it all became far too much for one individual, even two or three, to be the gatekeepers.

At my job at Cloudera, I have the privilege of working with Doug Cutting, who brought Lucene, Nutch, and Hadoop out through the Apache Foundation. Doug spent a bit of time with us talking about how Apache develops open source software and after a few weeks of work we produced a founding document for how MooTools projects will be developed and governed. This document, our “charter” is influenced heavily on Apache’s processes, though not as rigid as we don’t have their scale issues (yet!), but it’s had a remarkable influence on our development team. All the things I’ve listed in this lengthy post have sprung up as a result of this change. What started as a split (-core, to –more and -core) has now grown into a federation of projects that we govern as a group. And that federation of projects, that freedom to just make things happen, is now filtering out into our community with the Forge.

So when I look back on 2009 and ask myself what MooTools accomplished, I look at all the things we released, all the code we wrote, and all the new people contributing and think of those things as side effects. The thing that really happened in 2009, and it took all year to do it, was MooTools became a project owned by all of us. Valerio is still guiding it, but we’re no longer rate limited by his ability to pay attention to all we can create. And so projects move forward with their own teams, and plugins get released by enterprising individuals on the Forge, and users send us bug reports using MooShell to demonstrate the problem. All these things are spectacular. But what really happened was that (tiny at first) shift to belonging to everyone who contributes.

When I consider 2010, my mind is awash. In one year our team grew by several hundred percent and the number of projects being actively worked on did the same. Our community has better tools to share their work. 2010 is going to be less about MooTools itself and far more about what we, as a community, make with it. If 2009 was all about the MooTools developer community coming together and taking things to the next level, 2010 is going to be about our community of users, and the possibilities there are endless as far as I can tell.

So consider this a challenge, an open invitation. Surprise me.

Several weeks ago I had the pleasure of addressing a large crowd of Joomla developers at their conference in New York. The Joomla community is a fantastic bunch and the time I spent there talking about MooTools was perhaps some of the best I’ve had. The people present were very interested in MooTools and as well they should be. Joomla ships with MooTools built in and there is a large community of Joomla plugin developers who end up using MooTools because of it. Unfortunately, it still ships with 1.1.

A few months ago we, the MooTools team, decided to start focusing more on helping the Joomla community with their upgrade woes and began to re-address the compatibility code for 1.1 to 1.2 in earnest. When I spoke in New York to the Joomla team there and said that our goal was to have something to help them upgrade available by the end of the year, there was, dare I say, rawkus applause.

Well, here it is, Dec 31, and I’ve kept my promise. With the help of many of the MooTools Dev team (Chris Pojer, Darren Waddell, Nathan White, and others) we’ve finally got a compatibility layer that actually works and, more importantly, helpes you upgrade.

First, it logs warnings to the console (such as Firebug) and helps you find the places in your code that need updating. The script is not a compatibility layer as much as it is an upgrade helper. Hopefully this will help us retire 1.1 for good.

Secondly, easily half of the work, if not more, was spent creating a test environment for it. When MooTools 1.1 came out, we didn’t have a test framework to run against the code. When we released MooTools 1.2, we did, and it shipped with tests for most of the functionality included. So to make this compatibility layer, we had to go write tests for all of 1.1 (about 400 individual tests), then write the upgrade helper, compatibility code, and then run the tests from 1.1 against the 1.2 code base with the upgrade helper loaded. Creating the tests for a deprecated code base is no fun but I’m glad we did, otherwise the compatibility code would be in far worse shape.

But as a result of all that work by the MooTools Dev team, I can say that the helper script we released today is likely to cover about 95% of all the changes between 1.1 and 1.2; the other 5% being essentially un-hackable. Still, for most developers, it should cover all their work.

Seeing 2010 roll in with people dropping 1.1 is a great way to start it. If you’re still running 1.1, now is the time to move on.

You can find out more about the upgrade helper on the MooTools Blog.

For those of you who didn’t notice, MooTools released it’s long vaunted user generated plugin system today – The MooTools Forge [accompanying blog post].

First, this is just awesome news, and a big tip-o-the-hat to Guillermo Rauch, who wrote the app. For those of you who have been around our community for the last three years, the forge was always something we were “working on”. There have been many versions created over the last few years, and many other community created initiatives, that didn’t quite do what we wanted. Guillermo jumped in and built a great system, closely integrated with git, that, well, it kicks ass.

The Clientcide Libs

I’ve uploaded four plugins to the forge as part of the launch (Tabswapper, MenuSlider, HoverGroup, and dbug). I chose these because they were just the simplest plugins that I could release; they just don’t depend on much and they do one thing. Releasing Stickywin, for example, represents a much greater challenge.

One of the things the forge requires is individual git repositories for each plugin. This means that I have to split up my one big repo into lots of little ones, rewriting my download builder, docs engine, and other tools to make these things easy to manage. Consequently, moving things into the forge is likely to be a long and laborious task. For the moment, the four plugins I’ve released are just duplicate files of what’s in the single Clientcide repo, but eventually, once I’ve split them all up, I’ll remove all the files from the Clientcide repo and then include them all as submodules. For the time being, they will be duplicates, and releasing them will happen as I have time (which is not very much these days).

Good Times

I’m really excited about MooTools here at the end of 2009. I hope to make a longer post here looking back at all we’ve done this year, but certainly the Forge release will be a big highlight. 2010 has a lot of interesting things coming, I’m sure, but perhaps the thing I’m excited most to see is all the good things that the MooTools community creates and uploads to the Forge.

2009 has been a very busy year for MooTools and I’ve spent much of my time working on things other than Clientcide (like MooTools and Cloudera Desktop). MooTools More continues to grow – the latest releases (1.2.4.1 and 1.2.4.2) are full of all sorts of new features (over 50 plugins if you exclude all the date localization files), but you’ll probably recognize some old friends. Mask looks a lot like Modalizer, Spinner like Waiter, Form.Request like Fupdate, etc. That’s because in an effort to make MooTools more feature rich I and the other MooTools developers have been moving our work into the official project.

As such, Clientcide 2.2.0 is really a bug fix and compatibility release. There are very few new features but that’s because, in a way, all the new features are in MooTools More.

Here’s what’s new in 2.2.0:

• Modalizer is deprecated in favor of Mask
  • compat file is just the deprecated library; does not convert to Mask instances but rather just returns Modalizer as it always did
• StickyWin.Modal has new options for Mask; this is a breaking change
• Removed Element.Delegation (now in MooTools More; unchanged)
• Deprecated DollarE
• Deprecated DollarG
• Fupdate is now Form.Request in MooTools More; see compat files
  • Fupdate.Prompt is now Form.Request.Prompt
  • Fupdate.Append.Prompt is now Form.Request.Append.Prompt
  • Fupdate.AjaxPrompt is now Form.Request.AjaxPrompt
  • Fupdate.Append.AjaxPrompt are now Form.Request.Append.AjaxPrompt
• FormValidator got renamed in MooTools More to Form.Validator (the old name is preserved); updated FormValidator.Tips to Form.Validator.Tips (again, the old name still works).
• Collapsable is now Collapsible (that’s correct, I cannot spell). The old name is preserved (though the file name itself changed; update your build scripts).
• HtmlTable moved to MooTools More. No breaking changes, but the -more version has new features.
• Added Italian, Spanish translations for Simple Editor
• New StickyWin features
  • Z-index Ordering!
  • changing default value for allowMultipleByClass in StickyWin to true
  • Added setCaption method to StickyWin.UI
  • onUpdate event for StickyWin.Ajax
• added detach method to MooScroller
• dbug should now work in IE8
• offset options are now additive for pointy tip
• numerous bug fixes

For the most part compatibility is preserved. There’s that one breaking change in StickyWin.Modal where the options for the modal layer are different. I can only test this so much, so if you find areas where the compat files don’t provide coverages, please file a ticket.

I’ve also updated the download builder on Clientcide to use the now open source builder called MooTools Depender. Check out the blog post on MooTools.net if you want to know more about it. One cool thing about that you might find useful it is the ability to re-download a script you created with a given url. Now, when you download a script, the header includes a url to re-download the library with the same selections. This should make upgrading libraries a little easier.

I know what you’re thinking. It’s been 3 months since I last posted. My last post was all about how you can pay it back to MooTools, and since then I’ve been taking my own medicine.

MooTools 1.2.4 released today along with MooTools More 1.2.4.1. In this release, I’ve pushed 274 commits into the codebase (out of 398). There’s a LOT of new features in it, but some should look familiar.

For instance, my Waiter class is there, now called Spinner. And my Modalizer class is there, now called Mask. Oh, and Fupdate finally got a decent name; it’s Form.Request.

But there’s a lot of other stuff in there, and that’s not even the only place I’ve been cranking. There’s also a dependency loader for lazy loading MooTools scripts called Depender. There’s an all-JavaScript version and a server side implementation (that’s way faster; available as PHP or Python/Django).

And then there’s MooTools ART, which is awesome. It’s the beginning of a UI system for MooTools plugins.

Rather than go over all this stuff in detail though, I’ll just point you to the blog post I wrote over on the Cloudera Blog, which outlines all the cool stuff we’ve been building with all these new toys. There’s even a fancy screencast.

The Clientcide Libs and MooTools 1.2.4

But you’re probably asking yourself, what about the Clientcide plugins? Well, obviously I haven’t forgotten about them. First and foremost, it should be rather obvious by now that much of my work is now being pushed into MooTools instead of released here. It makes sense mostly, as the plugins get reviewed by the other MooTools dev team members and the result is always better. There are still a few plugins here that don’t belong in MooTools More and to that end I’ve been maintaining them, occasionally adding features.

The current version of Clientcide – 2.1.0 – should work fine with 1.2.4. The catch is that the upgrade path to switch to the new instances of the things that moved (for instance, Waiter moving to MooTools More and now being called Spinner), isn’t super duper easy. I do plan on authoring compat layers for some of these scripts, but not all of them.

What does this mean if you’re using these libraries? Well, you can keep using them. But I’m not going to maintain Waiter, for instance. I’ll focus on Spinner. Converting from one to the other should be relatively straight forward, but since they both work in the same space (because they have different names), you should be able to use either one or both.

Look for a release in a week or so of the next version of Clientcide for MooTools 1.2.4. There are several bug fixes and I will have some compatibility layers for some of the plugins. In the mean time, you should be able to update to 1.2.4 without any issues.

The Exceptions

The only exception here is that MooToosl More 1.2.4 includes a few plugins that are almost direct copies of what I have in 2.1.0, namely HtmlTable and Element.Delegation. If you’re using either of these scripts, you can just ditch the copy you get from me and use the ones in 1.2.4.

I get a lot of emails and tweets saying how awesome MooTools is. Sometimes people want to donate money or give us credit on their work in some way. Others ask how they can contribute. If you want to “pay it back” to MooTools, you can do so by:

• fixing bugs in our ticket list (lighthouse)
• submitting patches for bug fixes and features
• writing blog posts about interesting things you learn in the process of developing w/ MooTools
• get involved in our user forums and help those who understand it less than you
• release code you author that others might find useful

We don’t need credit in any overt manner. Your work should be all the credit we need. We would much rather see you involved in the community than any other form of gratitude.

I mentioned last week that I’ll be speaking at the Rich Web Experience in Orlando in December (1-4). Their site is finally online and you can see the lineup. It’s good stuff. I think this is shaping up to be the best JS conference so far this year, but of course I would say that.

Also, there’s this:

FYI…. Rich Web Experience 2009 is the first to offer all-inclusive pricing for attendees – registration, flight (continental U.S.) and 3 nights lodging! This offering allows individuals to get approval without having to jump through the traditional travel hoops!

Visit the site for more details and to see the schedule and stuff.

You can also catch them on twitter: @richweb2009.