gmmoris.Chékofif()

The oh too rarely updated blog of a freelance Front-End developer by Gidi Morris

[kofif-blog]

Bird Watcher JS

Birdwatcher is a little utility I made based on he brilliant Nicholas C. Zakas's blog post about his JavaScript error handling anti-pattern. You can read his post here: http://www.nczonline.net/blog/2009/04/28/javascript-error-handling-anti-pattern/

The basic idea behind the pattern is to spy on all our object in the application and detect when they misbehave and throw errors. The concept is much broader than that, obviously, and I suggest reading his post to understand the mind set, but the goal behind Birdwatcher (which is English slang for "a spy") was to make a light weight component that I cold use in my object instanciation factory in order to detect when problems occur but leave the actual implementation of what should be done about it to someone else - in my case it was my logging component but in your case if could be anything from UI messages for your user's benefit, to your analytics engine or to your permissions engine.

The repo is here: https://github.com/gmmorris/birdwatcherjs , have fun. ^_^

Basic usage

To use birdwatcher, all you have to do is tell birdwatcher to spy on an object and set the callback you wish to have called when an error is raised.

window.theKing = { giveUpThrone: function(){ throw new Error("The king will never be over thrown!"); } }; birdwatcher(window.theKing); birdwatcher.configuration({ onError:function(exception,method){ if(method == "giveUpThrone") { assassin.killKing(); } else { // some other error took place } } });

Advanced usage

Specifying different callbacks for a specific object

If the global Underscore variable isn't available (if you're using it in noConflict mode) then you can extend it manually by using the global '_pp' variable that is added by the script.

window.theKing = { giveUpThrone: function(){ throw new Error("The king will never be over thrown!"); } }; window.theAssassin = { killTheKing: function(){ this.injectPoison(window.theKing); }, killThePrince: function(){ throw new Error("No women and children!"); } }; birdwatcher(window.theKing); birdwatcher(window.theAssassin,{ onError:function(exception,method){ // notify the rebels that the assassin has failed // or rather when theAssassin.killThePrince() is called // presumably that wasn't prince Joffrey, because that was no child, that was a monster! } }); birdwatcher.configuration({ onError:function(exception,method){ // all other error which were raised if(method == "giveUpThrone") { //will be called when the king is challenged to give up the throne } } });

Preventing a rethrow

The default is to rethrow an error once it is caught, so that the normal error flow can be maintained even when an error is caught for special usages (such as logging). You cen prevent the rethrow from happening by setting the rethrow config to false. This can be used globally or at a single object's level.

birdwatcher.configuration({ rethrow:false });

[kofif-blog]

KanakanaK.js

KanakanaK is a small Javascript component I released recently as I had some free time to dig in my private repos and see what I can share.

It is a pure Javascript reverse CSS selector engine which calculates a DOM element's unique CSS selector on the current page. What this means is that it allows you to calculate a CSS selector which will give you a one-to-one way of referencing a specific node on the page.

This can be very useful if your own code runs on other people's websites as a tool (say as part of a chrome extension or something like that) and you need a way of referencing specific nodes on the page later after a refresh occurs.

The repo is here: https://github.com/gmmorris/kanakanak, fork to your heart's content.

Basic usage

To use KanakanaK to analyze an element and produce a unique CSS selector for it, all you have to do is call the global kanakanak variable as a function.

var myElement = document.getElementById("#myUniqueElement"); console.log(KanakanaK(myElement)); // prints "#myUniqueElement"

Configurations

The global KanakanaK variable has a method called "configuration" which can be used in two way: 1. Simple calling it will return an object which contains the current configuration values:

// calling KanakanaK.configuration(); //returns the object: { depth :3, specifictyThreshold:100, queryEngine :null, errorHandling :false, selectorMaxLength:512 }

Calling it with an object as a parameter allows you to replace the current configured values with new ones:

// calling KanakanaK.configuration({ depth :5 }); //returns the object: { depth :5, specifictyThreshold:100, queryEngine :null, errorHandling :false, selectorMaxLength:512 }

These are the different configuration options:

depth - How deep into the DOM hierarchy should KanakanaK go in order to reach a unique selector. This is a delicate game because the higher the number the more likely you are to reach a unique selector, but it also means a longer and more breakable one. Assuming you want to store this selector to use later, making it longer also means it is more likely to change and loose it's validity.

specifictyThreshold - A minimum specificty level. Once the parser reaches this level it starts verifying the selector after every method is called. This can cut down our execution time by avoiding needless parsing but can also hurt execution times by performing many verifications. Specificity is calculated based on the W3C spec: http://www.w3.org/TR/css3-selectors/#specificity

errorHandling - How to handle errors which occur during the analysis? The default is 'silent'. There are three possible values: 1) 'silent' which means nothing happens. The error is ignored and KanakanaK continues as usual. 2) 'verbose' which means the error will be thrown and has to be caught by the calling code. 3) A callback function which will be called with the error object as a first parameter and the element being parsed as second parameter.

selectorMaxLength - A maximum length for the CSS selector can be specified - if no specific selector can be found which is shorter than this length then it is treated as if no selector could be found. By default 512 characters.

Conflict

Just in case you also had the brilliant idea of using a variable called "KanakanaK", or you wish to move it off of the global object then you can use the noConflict method to receive a reference to the object and remove it from the window. Calling it will also revert the original value of window.KanakanaK which was there before loading the KanakanaK.js script (if there was one)

var myKanak = KanakanaK.noConflict(); myKanak(myElm);

[kofif-blog]

A SkipList for the soul

I've finally got my self to do some C++ work.

I haven't touched the language for over 8 year, which is a real pity, as I always feel so much more stimulated when I work in it.

So what have I done?

I have stared working on a big project I hope to release some time in the distant future (no, it won't be in the near future, as I am relearning C++ as I go, and this will take time) but I needed to get some of the ground work done first.

For my bigger projects I need a container which will allow me to maintain a string key based lookup table (sort of like a hashtable) but which also provides me with a sorted Key order (sort of like a linked list) and also contains pointers rather than the actual objects as I am managing my memory in a completely unrelated (to the container) place.

I couldn't find something suitable, so I made my own.

Skiplists

So what is a skip list? A skip list is a linked list which uses a sort of "highway" system to allow it to skip over multiple keys in order to reach the key it is looking for faster. It is a really cool data structure which I'd highly recommend getting familiar with.

Igor Ostrovsky wrote a very good explanation (here is Igor's blog) about this data structure, so I won't go into the nitty gritty, but I will tell you about my implementation. ... and I'll steal a little image he made (Igor, if you want me to remove the image, let me know)...  

SkipLists meet pointers and Hashtables

My implementation is written using good old fashioned C++, pure and clean, so if you are working in C++ and need a SkipList, I'd be honored if you chose mine.

It is built using a temple system, so all you have to do is create an instance of the list while specifying the type of object you intend on storing in it. Then you can start filling it with pointers to such objects along with string keys which will allow you to later retrieve them.

The rational behind inserting pointers and not the actual objects is that my implementation was built with a database's index in mind, rather than an STL container. This means that I wanted it to contain references to objects that I create within my system, but didn't want it to change them in any way.

Here is an example of how you can instantiate it and use it.

Start by defining the class of objects you wish to store.

class Wizard { private: string Name; bool IsEvil; public: Wizard(string name,bool isEvil) { this->Name = name; this->IsEvil = isEvil; } };

Now you want to create your list and add some objects to it.

LinkedPtrSkipList list(); Wizard* harry = new Wizard("Harry",false); Wizard* ron = new Wizard("Ron",false); Wizard* hermione = new Wizard("Hermione",false); Wizard* malfoy = new Wizard("Malfoy",true); Wizard* voldamort = new Wizard("HeWhoMustNotBeNamed",true); list.insert(harry); list.insert(ron); list.insert(hermione); list.insert(malfoy); list.insert(voldamort);

Now you can retrieve these objects, remove them etc.

Wizard* anotherPointerToharry = list.find("Harry"); Wizard* pointToNoOne = list.find("Nevile"); // pointToNoOne == NULL list.remove(ron);

So where can I get it?

Well, fork me! Here it is: LinkedPtrSkipList in libKofif 

So, is it production ready?

Well, that depends on how much you care about uptime. Seriously though, I only put it through basic unit tests and am now working on setting up a proper unit tests suite (I've never built anything large in C++ before, so as I said, its a learning process), but in the mean time if you want to play around with it and try it out - be my guest. 

I plan, in addition to testing it properly, on expanding it by adding iteration support and perhaps a few more useful properties as seen in other STL containers.

[kofif-blog]

I stand by my words

I'm about to ruffle some feathers.

Its an understatement to say that when you attack someone's beliefs they will naturally fight back. We all do it, and we all feel a need to do it, because the fact that something we have chosen to believe in at such an ingrained level has been questioned, can send us on a rampage of blind retaliation - no matter how much sense that person's words make.

We developers are no exception. No different than a religious zealot, we tend to make up our minds about a particular subject and  if someone questions that belief we lash out. This can be about a specific form of development management or a chosen  programing language.

By now you are probably wondering what it is I'm going on about.

Yesterday I met a colleague for a coffee break and we discussed possible directions for a new project. During this meeting, while discussing the possible frameworks to choose from, I made an offhand statement -  ASP.Net can't be scaled. I didn't even think what I was saying at the time, I was simply stating what seemed to me as fact.

My colleague looked up from his coffee and simply asked - Do you stand by those words?

I paused, and said I do, though I had to ask my self if that is true. I've spent the last day thinking this over and I realized that not only do I stand by those words I can explain why... this is no mere after thought.

So can ASP.Net scale

Of course it can.

Wait... whoah... but you just said?!?!

Yes, technically ASP.Net can scale. Technically. But it won't.

For the sake of argument lets make a stupid assumption. (Why stupid? Because it is wrong, but many entrepreneurs make this assumption when starting a new project, so we'll go with it).

When starting a new project you should choose one of the big names in web frameworks because these are proven in scale, and these are :Ruby on rails, DJango, MS.Net and Zend framework

What this admittedly silly but common assumption says is that the safest path is the trodden path and so we will choose one of four frameworks-  two free frameworks supported by a community and two comercial frameworks supported by companies.

I'll refrain from making a comparison between these frameworks, god knows that subject has been debated to death, but I'll make my point about ASP.Net specifically.

What are the problems that will in effect prevent ASP.Net from scaling

I am a firm believer in the following theories and stand by them:

Closed solutions. Large corporations, such as the Governmental body I worked for for seven long years, believe that they should take a path which can most easily be measured and presented in a Powerpoint presentation to it's board of directors. This in effect means that a comercial solution, supported by a comercial company (specifically in this case, Microsoft), is the safest choice - a big company will support any problem they find and will be more than happy to take millions of dollars in yearly consultation fees to make these problems go away. There are many problems with this work assumption but the key problems, the way I see them are them, are that the project will now be dependent on above stated company no matter what. This dependency can, and inevitably will, cost your project in both time and money which jeopardize it's success.

Comfort overhead. The strongest advocates of ASP.Net as a web framework talk about the ease of creating simple web pages quickly, with a very powerful IDE and an end to end solution for both your backend and frontend. What these advocates seem to miss is this - no matter how hard you want to believe that a framework can be an end-to-end solution, it cannot. Just as any developer  who has worked for longer than a month can attest, the term "patch" will inevitably find it's way into our code's documentation at some point and one patch leads to another. The overhead of a supposedly comfortable framework is that the more we hit a particular framework's wall of conformity (that invisible wall you hit when discovering that your chosen framework doesn't support what you want to achieve) manifests it's self in longer than normal development cycles and in unmaintainable code blocks. This overhead also manifests it's self in other measurable costs - comfortable frameworks, that claim to save your developer time by doing a lot of the heavy lifting for him, tend to prove heavier in the long run by creating memory intense or code heavy components (did someone say Viewstate?) that create a real overhead in performance and costs.

But none of this is new. These theories has been debated over and over again and many MS.Net advocates have retorted in many creative ways, many of them valid, and we still never endup with an actual answer.

This is where I'm about to ruffle some feathers

Theory number 3 is where things become really dirty.

Those who are now thinking that I have nothing new to say are compiling a love-hate response to my attempt to attack Microsoft's framework in an uninspiring fashion, but I'll ask them to wait a moment and read my next few lines, because they are the real reason for this article.

I have nothing against ASP.Net. I truly don't.

I have worked as a web developer for almost a decade, most of them spent in Visual studio's loving hands and I think Microsoft have built a very good framework.

And as many of the ASP.Net advocates will say, the framework can be scaled to support real scale. The framework supports a lot of very powerful HttpHandlers and HttpModules to control what is done by who and how he does it. You can easily bypass the bloated Page lifecycle and approach the project's development in the old fashioned scripting language style.

But that will never happen.

When I approach starting a new project and begin to think about future growth, developer recruiting and scaling problems I need to think about the limitations I may encounter on the way.

The sad truth about easy frameworks and closed solutions is that it dumbs down the developer community that uses them. I'm not claiming MS.Net developers are dumb. Not at all. I'm claiming that the developer community as a whole is less inclined to learning new approaches and new  techniques. I'm claiming that when your entire work experience is built on closed solutions, whose bare bones and muscle tissue are hidden from you and your only knowledge of infrastructure is that which is preceded by a "system.Something" prefix, you will find it too easy to slip into mediocracy.

This means, that if I am recruiting ASP.Net developers, even if the developer community is the biggest in the business (which in Israel, is true) the chances of me finding a high quality developer, who understand what an HttpModule is and how to use it, are slim to none. Especially if I need to compete with another one hundred CTOs searching for the same thing in the same pool of developers.

Inevitably I will find my self stuck with mediocre developers and big bills to Microsoft corp. for counciling fees.

My conclusion

This is why ASP.Net can only technically scale, because in effect it is too hard to find good quality developers who know how to bypass the framework's limitations once true scale is reached.

I am not saying these same problems don't exist with other frameworks, such as Rails or Django (Or web.py or Symphony etc.), but I do claim, that in the Israeli developer's landscape, the pool of developers working with Python or Ruby tends to be higher in quality and in understanding, when it comes to development challenges.

Where I get my heretic's ideas from

My theories are based both on experience and on blind belief, but I am not the only one who thinks this way, and I think that if you take a good long look at the Israeli Startup landscape you will find that our inventiveness and success is waining and that we need to take a good long look at the causes for this. I recommend reading the Humus manifesto which seems to affirm my theories from a more experienced thinker.

[kofif-blog]

Chrome misbehaves on submit

I came across an interesting little misbehavior of Chrome this morning.

So whats the dealio?

My QA guy complained that a particular feature was working on IE (yes, even IE 6) and FF but not on Chrome. The specific feature was a simple submit feature on a form which in turn sent a user to preview the changes he made before completing the changes he had asked for. 

So how did we look for the bug?

The first and obvious thing I checked was what the browser actually does- I pressed the offending button and found to my surprise that the form was being submitted to the server and even got a valid request. He just completely missed the fact that the preview button was pressed and treated it as the standard save request.

The next step was to debug on the server side and see what was going on in my precious server side code. I very quickly hit an obstacle -  I develop using Firefox and using the Zend toolbar to tell the browser to send it's request to the Zend debugger port which allows my Zend studio to latch onto the execution and allow be to debug the hell out of my code. Seeing as I now wanted to debug a request originating from Chrome - this was a problem. A quick search revealed that while Zend has not yet released an official Chrome extenssion, there is a developer's edition (you can get that here) which allows more or less the same functionality.

Once I got the extension installed I run my code, began to debug and discovered that Chrome didn't send the button click to the server - he registered a normal submit without registering the fact that I pressed the Preview button at all.

Some playing around taught me a new lesson on Chrome - it behaves slightly differently than other browsers with buttons in forms.

So whats the difference and how did you solve it?

Well, the difference is this - if you press a Submit button in a form it's name will be sent as part of the request in all browsers.

But, if you add a click event to your submit button and then manually submit the form - Chrome will not send the button's name as part of the request.

If to delve a little deaper into the problem, here is some code:

('mcPreview').addEvent('click', function ()  {                 var form = this.getParent('form');

                if(form != undefined)                 {                                 form.action = form.action + '#Preview';                                 form.submit();                 } });

As you can see, what I was doing was catching the buttons click event, changing my form's action (to send the user's focus to a specific part of the form after submission) and then calling the form's submission manually.

In Chrome, unlike Firefox or Internet Explorer, this code tells the form that the button is irrelevant to the request and doesn't send it.

The solution I came up with was to add a hidden field called action whose value I set according to the needed behavior (such as 'preview') and then in the server side I check that field's value instead of the button's.

[kofif-blog]

Setting up: Theres' an app for that

Moving on from our Operating system dilemmas we now need to think about what applications will suit our needs best to get things done.

These choices need to be made before we even start thinking about the specific application we intent on developing and obviously they are highly affected by the specific skills and knowledge we each have.

The following post will specify the applications I use personally and a quick brief on what this application does and what I use it for, broken down by the Operating system upon which each one of them has been installed:

Mac OS X

Lets start with the web browsers. Since this is a web development machine, one of the regrettable side  effects of not having to build a specific application for each platform you wish to support, is that you have to make sure your application supports all the different browsers those platforms use to access it. Confusing? You bet... to the point, this means I need to install every different browser I wish to support. While there are applications which emulate all the different browsers I have yet to find one which does the job properly and so, to be on the safe side, I use these browsers directly. 

Mozilla Firefox - My development browser of choice. Allows numerous plugins which can help in your development process. The plugins I always install are: GreaseMonkey, which allows me to manipulate Javascript to my heart's content (and being a monkey my self, I can relate to this Plugin on a personal level) and Firebug, which allows me to debug the DOM and use special tracing mechanisms in my applications (which I'll tell you about another time). 

Google Chrome - My surfing browser of choice. Fast, built well and fun to use.

Apple Safari - I never actually use it other than to make sure my creations work properly on it, but it is a rather popular browser.

Opera - Same as Safari - I never use it other than to check my creations as it has a nice market share, but it nice non the less.

Lets move on to the applications I use to get things done.

The mos basic tool in the developer's toolbox (yeah, not a cliche at all) is his IDE of choice. The place where he spends most of his time telling the computer what to do. So seeing as I now do PHP development with clientside development I need an environment to do all that stuff, so:

Though at work I use the Zend Studio IDE, it is costly and doesn't actually offer me much more than the open source options, so at home I use the Netbeans IDE, which is free, very good and did I mention free?

When it comes to working on the design side of client side I find CSSEdit to be a pleasure to use on a Mac. Robust, powerful and oh so beautiful, this tool allows me to inspect the client side code of other websites and learn from them or inspect my own development to allow powerful CSS debugging.. 

MAMP - An acronym for Macintosh, Apache, MySQL and PHP, this package installs, setsup and controls fully functional installations of Apache, MySQL and PHP on my Mac- in about 30 seconds. This allows me a fully working development environment to work on my mojo.

In the past, when I needed to find a colour for what ever I'd launch Photoshop or something similar - heavy and much more than I needed, I needed something lighter. Thats when I found ColourMod which is a widget for my dashboard and allows me to try different Hex and RGB values or search for a colour using a spectrum analyzer. 

No web development computer is complete without a way to move files around in a comfortable manner. Sure, I can SSH files around, but drag and drop is always more comfortable. For that I use Cyberduck, which is an open source and free FTP client with many tricks under it's sleeve.

Windows 7

Well, we've gone through the applications I use most on my OS X installation to get my work done. Lets take a look at what I use my Windows OS for (remembering that it is there mainly to allow me to do things that my OS X cannot do, such as run Microsoft applications).

Internet Explorer (7 & 8) are, sadly, amongst the most common browsers to surf our web applications (usually)... this is due to Microsoft's huge market share, and no much to do with the quality of the browser, but that meaningless - we need to support these browsers as well, and so I have these installed to allow me to properly check my applications to make sure they are running properly on different browsers.

Fiddler is the most exceptional little piece of software... frustratingly it isn't open sourced, nor is it multi platform and requires Windows, but I love it non the less. It allows us to analyze the traffic between any client machine and a web server- which helps us find bottle necks in our client side code, find heavy resources which hurt the user experience and find out many details which can help us improve our application's performance and responsiveness.

Visual Studio is an awesome environment for developing using the various Microsoft technologies. Though I am a big advocate for light scripting languages such as PHP or Ruby, I find Microsoft has created an incredible set of technologies, under the .Net name, which opens up a large new world for powerful web applications. At my past job I worked for 7 years on .Net applications and have a firm belief that there are many types of projects where .Net is very well suited to provide a steady framework. That said I also believe .Net has its limitations and many organizations fails to see that before hitting the barrier. I keep VS installed to allow me the freedom of choice between Microsoft.Net application development and all the rest.

Ubuntu

As you may have already guessed, I don't use Windows only for three applications, but those represent the main use. Ubuntu plays a similar role- I don't use it as my main OS (even though at my company we developers all use Ubuntu as our main workstations) but I find it comfortable to have an OS on my laptop which is similar to what my production servers run. The majority of projects I have come across in my line of work have been either Windows Server machines (for ASP.Net projects) or Linux machines (usually Red hat or Centos) and having the ability to test things for these operating system on my laptop is very comfortable.

I can also test the way Firefox and Chrome render on Linux, but I hardly do that, since they appear to render in the exact same way and most problems arise with third party components such as Adobe Flash.

[kofif-blog]

Setting up: The workhorse

Complex.

Complex is how I would describe the process of setting up my workhorse has shown itself to be. I'm referring to getting all three operating systems to work together.

Just to remind you what our goal is:

Install OSX, Ubuntu and Windows 7 in parallel on the same machine.

Be able to boot into each one of these independently.

Be able to virtually boot Ubuntu and Windows 7 from within OSX.

This is no simple task, my friends.

What didn't work

Well, most things.

My first attempt went as so:

Installed OSX from scratch.

Partitioned the harddrive to make space for Windows and Ubuntu.

Installed Windows on one of the new partitions.

Installed Ubuntu on the other partition.

Tried to load OSs.

The result? Windows broke- simply wouldn't boot. Ubuntu loads just fine... but has trouble accessing harddrive. OSX - works flawlessly.

Failed miserably.

My second attempt went so:

Installed OSX from scratch.

Partitioned the harddrive using bootcamp.

Installed Windows on bootcamp.

Resized the Bootcamp partition and added new partition for Ubuntu.

Installed Ubuntu on the other partition.

Tried to load OSs.

The result? Windows loads via Grub (a Linux bootloader). Ubuntu loads just fine via Grub. OSX - works flawlessly.

Great! So what do we do? Try to virtualize Windows using Parallels - fails miserably. Simply can't do it.

Hence, the updates result?

Failed miserably.

I won't bore you with a list of my failed attempts, but just to give you an idea:

I tried reversing the order of the OS installations.

I tried Virtualizing before installing a second OS.

I tried using rEFIt to act as a bootloader - had no effect.

I tried partitioning using different kinds of ERDs.

Etc...

Since every time I tried one of these permutation it involved reinstalling, from scratch, at least two Operating systems, you can imagine how long this took (and how frustraiting it was!). :)

Anyway, on to The solution Finally, after following some ideas thrown my way by members of Superuser.com, I found the solution to my problem.

Since I figured out that Grub was the main culprit preventing me from virtualizing my Windows and Linux installations I had to find another way of bootstrapping Windows and Linux alongside OSX which wouldn't prevent Parallels from virtualizing my Bootcamp partition.

Wubi to the rescue.

Quoting the Wubi website:

Wubi is an officially supported Ubuntu installer for Windows users

that can bring you to the Linux world with a single click. Wubi allows you to install and uninstall Ubuntu as any other Windows application, in a simple and safe way.

Basically, the main difference between Wubi and the normal installation of Ubuntu, is that when you format your harddrive and install Ubuntu directly onto it, the installer will also install Grub in order to allow you to dual boot your computer with other OSs.

Wubi, on the other hand, will install itself via your Windows installation (It is simple an executable you run from within Windows) and add itself to Window's standard Bootloader.

Effectively, this mean, that your windows installation stays normal and intact and you can simply boot into Ubuntu instead of it if you wish... and the nicest feature- if you are sick and tired of Ubuntu's quirks (such as not being anle to press a button in Java application without using the Return key...) you can simpy use Window's Add/Remove programs to uninstall it.

For me, this is invaluable, since Parallels Desktop now sees my Windows installation properly and can virtualize it properly. The real bonus is, that since Ubuntu runs through the Windows Bootloader, all I need to do to virtualize it as well is run my normal Bootcamp virtual machine and during the boot process I can select to boot Ubuntu.

So, to recap the solution (follow these instructions properly and you will be fine):

Format your whole Macbook Pro's hard drive.

Install OS X: Snow leopard on your machine.

Use Bootcamp to partition your harddrive for Windows 7 (take into account that the space you allocate will serve both Windows and Ubuntu, so make a partition large enough for both).

Install Windows 7.

Boot back into OSX and use Parallels to virtualize it.

Once you can virtually boot Windows, use Wubi to partition and install Ubuntu under your Windows installation.

Boot which ever OS you want, or virtually boot Windows 7 or Ubuntu via Parallels under your OSX installation.

Bobs' your uncle.

And by using this setup I have met all three of the forementioned goals from my previous post...

Now that the infrastructure is done, I need to install software on all three Operating Systems to allow me smooth sailing through the world of Multi platform Internet Development and Design.

Oh Joy. :p

[kofif-blog]

Setting up: Intro

Freedom!

One of the most dramatic changes this new job brings with it is the fact that I am no longer a Windows shackled robot. In my job of the past 7 years, I have worked exclusively on Windows machines, using Microsoft technologies and techniques to create boring, unimaginative and repetitive Knowledge management systems for a large governmental body.

But now, things have changed. I have broken loose of my suited-up, strict and uninspiring shackles and have joined a young, energetic and fun company whose goal in life is to keep people entertained... and find a way to make a living off of it, in the process.

The first stage in this new adventure - setting up my workhorse. My laptop, which has always been my one source of sunshine in an oppressive cave - is now to adopt a new role. My Apple Mackbook Pro- a sleek, handsome and ooh-look-so-bootiful work machine, has travelled with me around Israel and all the way to Mexico (and back) and has never been a work laptop, it has been a joy laptop. But now, it is to take on the role of my new - Totally awesome lean mean dev machine.

The architecture

The plan is simple: since my OS of choice is OSX (Snow Leopard) and my new position requires Linux (Ubuntu) an OSX-Ubuntu dual boot seems like the ideal solution.

But sadly, that doesn't actually fit the bill. You see, being an Internet developer, I am faced with the Multiple platform problem. The end user, who reaches my product, could be coming in on any platform he sees fit - Firefox on Fedora, Saffari on OSX, Internet Explorer on Windows Vista (yuck!) or any permutation between those.

This means I must have the tools to do the following:

Develop multiple web technologies on multiple platforms

Allow me to test my own creations on every relevant browser\platform

Allow me normal daily use

In order to provide solutions for all three prerequisites specified above I have defined the following guidelines:

I must have all three operating systems accessible- OSX Snow Leopard, Windows 7 and Ubuntu.

I must be able to boot into each of them individual - Since all three represent development machines, loading them as virtual machines isn't a viable option.

I must be able to load Windows & Ubuntu virtually via Parallels Desktop, since that way I can use my Windows apps and Ubuntu apps in OSX, which will serve as my main OS.

So how do I set this up? Read on to find out.

Freedom rules!