A couple of days ago I was introduced to a new data structure I did not know before. It’s called a skip list. It’s a sorted linked list, where the nodes are linked at different levels, forming increasingly sparse chains, each of which skips more nodes, and therefore is faster to traverse, allowing search algorithms to hop over parts of the list when searching. That’s why they’re called skip lists.

The concept was shown to me during a job interview. They briefly introduced me to the concept of skip lists using an image. It is said that an image is worth a thousand words, and in this case the assertion holds:

After I finished the interview I went on to find out more about skip lists, and I was amazed on how I did not stumbled upon them before. Immediately my instinct was to try out the algorithm I came up with during my interview, and when I saw it worked, I ended up writing a whole new SortedMap subclass in Java, implemented with skip lists.

It is surprisingly amazing how easy it is to write the routines for all the basic operations on skip lists, like adding, updating, removing and querying on them, specially when you see it described as a probabilistic data structure, which you have to admit that looks scary at first glance.

Performance

Well some guys did something very similar. They took an iPad for a ride in a Zero-G parabolic flight, similar to the well-known NASA’s Vomit Comet. Can you guess what happened to the iPad? Surprise!

Maybe not as serious as the problem that Jim Lovell and his crewmates had on Apollo 13, but a problem indeed. I was really not surprised at all, since I expected no more, but it’s nice to see that someone actually gave it a try.

Personally I always considered it the best browser ever, although with great grief I had to hand over that title to Chrome, which has everything really worth taking from Opera, plus several improvements that were hard to ignore. It was the first browser able to claim being nearly as fast as the Norwegian big O, plus sandboxing, well-achieved extensions framework, uncluttered interface and a great attention to detail from its designers.

But I didn’t throw away Opera for this. Today I took a glimpse of what’s expected for their upcoming 11th edition, and I have to say I was impressed but not surprised at all: I expected no less from the best browser vendor in the world.

Opera 11 beta brings up several improvements over past versions. Keeping up with the competition is the least they could, and they seem to be doing really well in terms of features: better HTML5 support, smaller installer, better performance, safer browsing, secure and informative address field, private navigation mode, and a great new overdue addition which I haven’t yet explored in its entirety: extensions.

Something really new

But all these only puts it on par with competitors, feature-wise. What’s really hard these days is to come up with new stuff; unprecedented functionality on an era where everything seems to be already invented. That’s what’s truly amazing of a new version of any software, particularly web browsers. And Opera has always been one of the bests at this.

This 11th version takes tabs to a next level: tab groups or stacks. Users can manually group tasks on stacks that can be expanded or collapsed at will, cleaning up space and clutter on the tab bar. For heavy-weight Internet users like me this is a great idea. Watch the video at youtube to see how it works, or download the browser and try it yourself.

Managing stacks from thumbnails preview

This feature gets even better when combined with floating thumbnail previews. When you hover the mouse pointer over a collapsed stack, a large floating view appears with a grid of small thumbnail views of each tab in the stack. You can click on the thumbnails in the floating view to switch tabs without having to expand the stack. Watch it working in the following video.

What I really miss and I hope it’ll be included in the final non-beta version, is the ability to specify in the preferences that when middle-clicking links to open in new tabs, these new tabs get automatically stacked with the tab that spawned them. This should of course be optional and not enabled by default, or even made available only with some modifier key while middle-clicking.

The verdict

This is still a beta release with some rough edges, specially a little prone to instability and crashing. But it’s good to know Opera is still heading in a good direction. They lost too much time with stuff like BitTorrent support and Opera Unite, instead of focusing in the browsing experience. Hopefully this browser will someday get the attention it deserves.

What do you think about Opera 11 and tab stacks? Do you think this feature will make its way to other browsers in the future?

I was most recently getting along with PlainText, a note-taking iPhone app that stores your notes in a folder in your Dropbox account. It’s a very simple yet powerful concept, specially for those of us who already use Dropbox. Instead of having yet another syncing environment for notes, they already take advantage of the Dropbox user base out there.

The interface is simple and visually appealing, yet out of the ordinary. The notes are always in sync between the iPhone and the computers thanks to Dropbox, and I can get to organize my notes into folders. No more nasty notes from the default iPhone notes app getting into my mail (whoever thought that notes were related to mail is insane).

But then the simple got even simpler…

Today I stumbled on a simpler and better alternative: SimpleNote. It’s an online web service with an iPhone app, an API, and a set of third-party supporting apps and tools that complete the picture.

Great features

The killer features over PlainText are tagging, versioning and sharing. Yes, you read it right.

With SimpleNote you organize your notes with tags, instead of placing them on folders. You can associate any number of tags to a single note, and you can also list all notes regardless of tagging. It’s of course a bit unfair to make this comparison against PlainText, since its Dropbox storage paradigm would not allow for anything else than folders.

You can also share notes with other people. I did not found a way to do this on the web app, although there should be. In the iPhone app you can activate sharing, and it’ll give you a URL, plus the possibility to send an email with an invitation to the share note. I haven’t tested this though because I still have no one on SimpleNote to share with.

Also worth mentioning are its versioning capabilities. You can check past versions of your notes, up to a limit of 10 versions in the past. There’s a premium non-free plan for a mere $12 a year (that’s $1 per month) which extends this limit to 30, plus extra features such as no ads, better support, creating notes by email and more.

On the desktop side of the equation

On the desktop side there’s no “official” app for SimpleNote, but on the downloads page I found a couple of third-party client applications for Windows and Mac that I am really liking. On the Mac side we have JustNotes, which syncs with the online service and supports tagging and searching. On the Windows side I was not expecting much, but it turns out that ResophNotes is a pretty good client with email, printing and Markdown support. The downside is that it does not support tagging. It did not support tagging until recently, so be sure to have the latest version.

There are some other options for both platforms, specially for the Mac, and there are options for Android as well. There’s even an Emacs package and extensions for Firefox and Chrome!

What other alternative services do you use for note-taking on the iPhone with desktop-sync support?

You gotta love Jim Parsons on this scene. For me it’s the funniest scene on the whole series so far. It makes me laugh every time I see it. I hope you enjoy it too.

Update: Unfortunately the Youtube video above is not embeddable, so it cannot be played within this site Simply click on it to be given the option to open it on Youtube instead.

Luckily there are lots of different choices of password managers, softwares that will help you keep a secure database of all this information, sometimes even in a way that they help you collect your passwords as you use them in your browser, autofill login forms, and even generate new random passwords for you.

This is all great, but I am not convinced. Random passwords generated and filled into forms, all done by a software tool? I end up not really knowing my passwords. What if the tool ever fails, or its database gets corrupted? What if my hard drive melts and after the fix I get a clean install with no passwords database? I also do not like the idea of passwords being stored digitally, not even encrypted.

So I ended up with a small password managing system that I designed myself. With it you can still use a conventional password manager, but in a slightly different way.

Grab some passwords…

I have a small set of around nine or ten different passwords that I use for all the services and accounts I have access to. These passwords range across all the spectrum of difficulty and strength.

Then I distribute this set of passwords more or less evenly across the whole spectrum of services and accounts that I have access to. I use strong passwords for the most precious and important services; the dumb passwords for unimportant stuff, like all the garbage sites that require membership for even the dumbest thing; and I leave the medium-range passwords for most of the average-but-important services that most people, including me, use these days.

…but store keywords instead

For this password system to work you really ought to remember the actual passwords yourself. The password managing system is mostly just to remember what passwords are used for what access, not to remember the actual passwords. You can keep these associations between user accounts and passwords stored in several different ways, so it’s pretty easy to use, not needing any specific software. You can use an actual password manager, or you can use an Excel spreadsheet or a simple text file, just as I do. The medium is really up to you.

There are a couple of rules when setting up this associations list

1. Do not store the actual passwords in the associations, just small bits of text that loosely resemble, reminds you, or have any obscure-for-anybody-else relation to the password. It can be a small bit of the password itself, perhaps a bit tweaked, reversed, or with some phonetical changes. Or it can be something textually unrelated to the password, but having some association with it that only you can devise.
2. Each password should have at least two different keywords that you’ll associate with it, and not just one. This ensures that if somebody ever gets access to this associations, it won’t be easy to determine what services use the same password.

An example

I’ll give you an example with my own passwords and keywords (not really). We’ll work with a set of six passwords with various levels of strength. Notice in the stronger cases the use of L33T 5PEAK and other tweaks.

As you can see, most keywords by itself do not reveal the password, not even remotely. And even if someone ever guesses the relation, the password itself is usually more than the part that links to the keyword, with the further complication of tweaks in the spelling, the use leetspeak and similar encodings, etc.

But remember that you do not really write down the table above. What you really do is write down a list of services and accounts, and the keyword that associates a password with it, instead of the actual password. During the creation of this list below is when you really think about keywords associated to your passwords.

You can now store this table in whatever format you prefer, and take whatever measures to not loose it. You should also store it in a secure location, but in a way that you can access it whenever you need to lookup for something, or to modify it. Dropbox and Evernote are good choices. In Dropbox you can store it in a password protected file as well, if you think you can use some extra security.

But…

Sure this system has some drawbacks. For instance, some people might argue that some password managers also fill out login forms for you when integrated to the web browser. So even if you do use one of these password managers to store just the keywords and not the actual passwords, then the autofill-login-form feature becomes useless. But hey, who access content on the Internet via web browsers anymore?

What tools or systems do you use to manage your ever-increasing list of passwords and user accounts?

Objective-C provides most of the benefits of C++, plus some more benefits that C++ does not provide at all. And I am not talking only about Simula vs. Smalltalk philosophical differences here, but about stuff that C++ could have actually had without breaking its Simula-like philosophy. Some of the features/reasons I will enumerate below fall under this category.

Declared properties

Objective-C provides a simple way to declare properties that define and describe the state of an object, as well as to implement the way we access these properties.

Properties ensure that we adhere to the principle of encapsulation, by which an object controls the way that other objects have access to query its own state. In Objective-C all instance variables are private to an object, and its values can only be accessed via properties defined in the object’s class.

Properties also play a big role in some of the core features of the runtime library, such as KVC, KVO, Model Objects, etc. I wish C++ would have had this one.

Protocols

Protocols are similar to interfaces in Java. When I started learning Java over twelve years ago, no one made a reference to Objective-C protocols as a similar feature of an existing language, even though protocols were introduced to Objective-C several years before Java appeared.

A protocol defines an abstract interface that classes can adhere to. This interface consists of a set of method definitions that implementing classes should provide.

As a plus, compared to Java interfaces, protocols can define some or all of its methods to be optional, so that implementing classes can opt to provide implementations for the optional methods or not. In this case when sending a message to invoke an optional method on an object that conforms to the protocol, the caller should ensure first, via reflection, that the object is able to respond to the message.

Compared to C++, protocols provide a way to circumvent the lack of multiple inheritance, although their use in the standard Cocoa library is mostly not related to this.

Categories

Categories allow you to add methods to an existing class, but without subclassing. You can do this even with classes from the runtime library or classes to which you do not have the source. This is my favorite language feature. Ever since I became addicted to Ruby I believe I cannot stand a language that does not allow me to do this.

When you first think about it, it makes no sense. Subclassing should be fine. But what if you need to add a new functionality to an existing hierarchy of classes? The best example comes not from Objective-C itself, but from Ruby and the Ruby on Rails framework: Rails would not have been possible without Ruby having this feature. Another good example within the Objective-C realm is the Three20 iPhone programming library, that would have been harder to implement and maybe not too feature-wide if it were not for the language having this feature.

A more mundane but valid use for categories is to split the implementation of a class into separate source files, be it for organizational purposes, or because the class interface and implementation is too extensive and splitting it up makes it easier to understand and maintain, etc.

This is one of the features I believe that C++ could have had without breaking its philosophy, and it would have been great.

Compatibility with C (and C++)

Unlike C++, Objective-C is a proper superset of C, meaning that any C program or file is also Objective-C code. There’s no guideline of what to do and what not to do when using C code in Objective-C. The most evident advantage is that Objective-C programs can use any existing C library without the need for wrappers. Some notable examples of libraries that are used in iPhone programming are sqlite and OpenGL.

It should also be mentioned at this point as a side note, that Objective-C can be combined with C++ as well, although they are still different languages. In this case there’s a list of requirements or restrictions to observe. For more information refer to the official documentation about this feature.

Simpler, more flexible, and easier to master

This one is a subjective reason, but I stand by it.

I have mastered more of Objective-C in six months than of C++ in years. I have also been more confortable learning it. I remember learning C++ as a necessary evil, and by contrast, I am happy that iPhone programming led me to this wonderful programming language.

The language is far from perfect, and there are certainly a few features from other languages that I miss a little bit. My perception of it has changed over time though. When you first start to use it, it appears to be cumbersome and bizarre. But sooner rather than later you start to love it (unless you are a C++ fanboy, of course).

So what do you think about Objective-C? Are there any other features that you love or hate about it? Feel free to share you opinion without starting a war.

Well, not exactly. If you have a real device (and you should anyway, if you are developing for the platform) and this device has lots of contacts in it, then you can copy the internal address book files into the simulator’s file system in your Mac. I actually got the tip from this blog post after searching on Google a little bit, but it did not worked for me exactly that way. I’ll let you know why.

First you have to extract the device’s address book sqlite files. There are two files actually, and according to their names, one is for the actual contacts, and the other one stores the contact images. Both files are located in /var/mobile/Library/AddressBook/ inyour iPhone, so if you have a jailbroken device, then it is as simple as copying through ssh…

scp mobile@iphone:Library/AddressBook/*  ~/Library/Application\ Support/iPhone\ Simulator/User/Library/AddressBook

If your device is not jailbroken then it is a bit more complicated. The solution involves extracting the sqlite files from the backups of your device created by iTunes in your Mac whenever you sync. The blog post linked above has a nice script for doing this, although I have not tested it.

But this did not work for me right away. I copied the files into the specified directory and started the simulation just to see the same empty list. So I wandered a bit through the vicinities of that location and I realized that along with the User directory inside ~/Library/Application Support/iPhone Simulator, there are a few more directories whose names correspond to the versions of the SDK installed in the simulator. Since I am running the simulator on iOS 4.0.2, then I copied the files into this directory structure instead, like below

scp mobile@iphone:Library/AddressBook/*  ~/Library/Application\ Support/iPhone\ Simulator/4.0.2/Library/AddressBook

So if the User directory don’t work for you either, then try looking for the one corresponding to the SDK version that you are running in the simulator.

So besides the obvious hiding of sections not relevant on a printed media, like the sidebar, comments form, etc., I started to look at what other improvements could be made for a printed version of a web page. I readily searched for and turned to an article I recall having read some time ago in A List Apart about exactly this matter: Going to Print.

Following the article’s suggestions I started customizing margins, backgrounds, font type and size, etc. Then I stumbled upon a very tricky issue they mention: what to do with links? They are clearly of no direct use in printed media since users can’t click on a paper, but on the other hand it’s a pity that the extra information conveyed in a link (namely the referenced URL) gets lost.

CSS selectors to the rescue

The article suggests using an advanced CSS2 technique. Using special selectors we can append after every link, the URL it refers to enclosed in parentheses. This will only be visible using modern and fully CSS2-compliant browsers (Yes, I am staring at you, Internet Explorer).

#content .post a:after {
  content: " (" attr(href) ") ";
  font-size: 90%;
}

The above rule states that after any anchor tag within the post, the href attribute value should be added in parenthesis. It also additionally puts the link in a slightly smaller font so that the link text gets more attention.

However, when I actually tried this I didn’t like it. The resulting printed view is too cluttered, specially if you have a sizable amount of links in your post, or if some of them reference URLs that are too long and hard to read. It would be really nice if one could selectively apply this printed styling to certain links only.

An apparent solution would be to apply a certain CSS class to the links we want to have this behavior, and then specify in our style sheet that the rule above should only apply to links with this CSS class.

#content .post a.print-href:after {
  content: " (" attr(href) ") ";
  font-size: 90%;
}

Note the .print-href class in the CSS selector. We now have to apply this class to all the links in a post that we want to be printed along with the referenced URL.

The problem is that there’s no straightforward way of doing this in WordPress’ post editor. The link popup allows you to specify a style class to be selected from a predefined set of wordpress-specific classes, but not your own.

Plugins to save the day

The short story about how to solve this, involves installing the TinyMCE Advanced plugin in WordPress, which enhances the post editor in several ways. One of the several enhancements is that the link popup is larger and has more options, even a few separate tabs grouping lots of different options. The last tab, the one labeled Advanced, allows you to arbitrarily specify the value for several possible attributes of the anchor tag, including the class attribute.

Notice the classes text field in the picture above, where it is already filled with the class name. The actual class name doesn’t matter, as long as you use the same class name in both your links and the CSS rule in your stylesheet.

So now, if you print this post, you’ll notice how some links in it get printed along with the referenced web page, and others don’t. I hope this hint may helpful to anybody out there.

Do you know any other helpful hints about making web pages and blog posts more printer-friendly?

Since I don’t have $14 to spare on this app, and I also do not really think it’s worth it, I went on to find some alternative softwares for this. This wonderful website alternativeto.net allows you to look for software applications on certain platforms and find out about some other softwares that could serve as alternatives or equivalents, either in the same platform but free, or perhaps because you are looking for an alternative to switch or use in another platform.

In this case I wanted to find an alternative to Divvy also for the Mac, but free as in beer. I didn’t really care that much if it was open source, although it would be a plus. Thankfully I found ShiftIt, wich is not only free and open source, but also even better than Divvy in my opinion.

ShiftIt, just as Divvy, sits unobtrusively in the menu bar. It would be nicer if it had an icon instead of text, but that’s a minor complain. What makes it great is that it already comes with sensible default shortcuts for positioning windows in logical default locations, such as covering the left or right halves of the screen, or top, or bottom. Plus covering the four screen areas formed by dividing it in four parts: top left, top right, bottom left and bottom right. And of course, two more options for full screen and center a window respectively.

When I saw this I immediately uninstalled Divvy. I found in ShiftIt preconfigured what I was manually configuring in Divvy just a few minutes ago. Exactly those very options, perhaps with some similar shortcuts I invented. Somebody would say that ShiftIt should allow you to customize the shortcuts and/or allow you to add some other configurations, but I guess they just want to keep it simple, and as simple as it is, it is still useful and able to replace Divvy 90% of the time.

I also found out about other similar alternatives, like SizeUp, Cinch, etc. But I only really tried the free one, as this was my goal. But maybe you want to try them all and stick with what you think is the best. And don’t be shy about making any suggestions here, either free software or not.

Do you know about any alternatives to Divvy and ShiftIt, be it for the Mac or not? Maybe with better or different features?