CAPTCHA Fail

I can understand the need to filter humans from machines on the Internet, but we need to do a much better job. It seems like the most common captcha software I run into these days is reCAPTCHA. The system has some noble goals - specifically, users who submit captchas are actually helping digitize books. However, here's the problem:

What the hell does that say? The first word is obviously "the", the second? "Claccupr?". Perhaps I have bad eyesight, but I often get these wrong. Here's an alternative proposal:

Instead of asking humans to perform text recognition (something computers can already do reasonably well), ask them to do something that comes naturally to humans, but not to computers. How about asking the user to pick which image they might find at a beach. Or ask the user to pick the image that represents the largest object in real life. Use context that humans understand instinctively, but computers do not.

Any ideas? Someone should build it!

Webkit / Konqueror issue raised again

Just thought I'd point out that I'm not the only one who would like Konqueror to use webkit rather than KHTML:

WebKit in Konqueror


...It's a pity the comments are 90% flame, and 10% content.

Henry lives on

After complaining about the poor state of the web browser in a KDE platform, I have to report with mixed emotions that I've bitten the bullet and installed firefox. I'm not a huge fan of firefox - yes, it's open source, and seems to work fairly well, but it's also slow and a huge resource hog.

Who here remembers when firefox first came out? It was supposed to be a stripped down version of the mozilla web browser. The idea was that by removing the mail client, IRC chat application, and god knows how many other applications we'd end up with a smaller, faster, lighter browser. To some extent it worked. However, I'm starting to wonder if they'd have been better starting from scratch.

I challenge anyone reading this to use Chrome for windows for a week and then switch back to Firefox for good - I guarantee you you'll be pulling your hair out within a week; firefox is slow! I always assumed that the reason my browsing experience was so poor was down to my slow Internet connection, but it turns out that a fair amount of the delay is the browser.

So I have firefox - the GTK theme KDE installs looks awful, and several web sites look rubbish, but at least I can check my email...

Well, that's it for now. More to come soon (and this time I'll lose the shakespearean titles).

My Kingdom for a Browser!

This post is set to be one of the most painful entries I have ever written on this weblog. Not because the subject matter is particularly difficult, but because the technology has let me down.

The story starts with me upgrading my laptop to Kubuntu 8.10. It's been out for a while, and I'm a big fan of KDE 4, but I hadn't had a sufficiently quiet weekend in which to take the plunge. I was previously running Kubuntu 8.04, so I could have just downloaded the latest packages, but I wanted to start from scratch, for a couple of reasons.:

1. I wanted to remove all the rubbish that I had installed over the last six months. I frequently download and install applications, only to find that they're not quite what I want. I rarely uninstall them, so over time my lhard disk fills with cruft.
   
   
2. I wanted to wipe away all the stale config, especially as my window manager would be changing from KDE 3.x to KDE4. Besides, there's a certain pleasure to be derived from configuring a brand new KDE installation.
   

The install was a breeze, and for the first time ever all my laptop hardware was detected and configured correctly without any hacking on my part - even the weird web-cam, which doesn't even work in Windows XP. Life was good, until I went to browse the Internet.

KDE ships with Konqueror as it's default web browser. As far as web browsers go it's fairly nice - It lacks the large "Add-Ons" repository that Firefox has, but many of the plugins I can't live without when using Firefox are included as standard in Konqueror.



Konqueror is more than just a web browser though - the integration between konqueror and the rest of KDE is truly stunning (as an aside: this is why I prefer KDE over other desktops. Technologies like KPart and DBus are the future of desktop applications, and KDE is leading the charge in this area). As an example, if you want to search google for something, but don't have your browser window open, what can you do? Easy! just press Alt + F2 to open the "Run Command" dialog, and type "gg: " followed by your desired keywords. Hit enter and you'll launch Konqueror with the google results right there waiting for you.


Konqueror also has extensive protocol support. For example, SCP and SFTP are supported by default. Try typing something like "fish://user@host" - konqueror will as for the user password, and will then behave like a file browser for the remote machine.

These two examples hardly scratch the surface of what Konqueror can do. However - there are some very serious problems with it. Using GMail with Konqueror is torturous. First Google will give you the plain-old-HTML-only mode, since Konqueror isn't officially supported. Then, if you ask for the full version anyway you get all sorts of weirdness - and a completely unusable inbox. The solution seems to be to set the user agent to Safari 2.0, but even then my inbox seems to be incredibly slow.

Members of the KDE community have pointed out that GMail plays fast-and-loose with web standards, so it's understandable that Konqueror misses a few tricks. The Google engineers must have tested the javascript enhanced version of GMail with the most popular browsers, and left Konqueror out in the cold - and fair enough. However, the KDE developers are missing the point: no matter how good their browser is technically - no matter how standards compliant it is, it simply does not work for me - the user. I now have a browser that I cannot use to check my email (no, using the HTML-only version is not an option).

So what are the alternatives?

Before I upgraded Kubuntu I had Firefox installed. However, when I went to install it, I nearly had a heart attack. In order to install Firefox, I had to install 63 other packages - most of them gnome or GTK packages. The reason for this is simple: Firefox uses the GTK toolkit to provide a UI. I knew this already, but this early on in my new Kubuntu install I wasn't about to pollute my OS install with GTK packages.


What can I do? There are a few other options available to me:

There's been talk of a Firefox port to Qt. However, nothing usable has materialised yet, so that's off the cards.

There's the Arora browser - this is a Qt browser running the Webkit engine (which is included as standard in later Qt distributions). A quick install told me what I needed to know: also not really usable as my default browser.

Finally there's Google's offering: Chromium. However, this has not yet been ported to Linux.

So what's the underlying cause of my troubles? Without hacking the code directly, I have no idea. Perhaps this is part of the KHTML vs Webkit debacle - There's a good article outlining the whole issue here, but I'd like to quote a couple of paragraphs:

So, what's the situation? Well, it appears that KHTML will remain the web rendering engine for Konqueror going into KDE 4.0, and that it could be changed to qtWebkit as of KDE 4.1. That does not seem to be officially settled, so much as the most likely scenario. It appears that the KHTML team seems hesitant about the proposition, while many KDE developers and users alike have expressed a very receptive attitude toward seeing Konqueror user qtWebkit. And Rusin made clear to a reader that he believes the KHTML team should continue their work as long as they like.

The challenge is that Webkit, which comes from Apple, is widely tested, and is thus known to work well with a large number of websites. KHTML is not as widely tested, and, for example, GMail doesn't work well with Konqueror. Many Konqueror fans have expressed regret at having to keep Firefox around just for sites like GMail, that don't recognize KHTML. Using Webkit would solve these problems, enabling many users to stick to one browser.

In other words: "The developers are dragging their feet to implement a fix that would arguably make Konqueror a better browser". Of course, the developers involved are free to do as they please with their code, but they're dragging down the rest of the KDE platform - I now have to have multiple browsers installed to do the most basic of day-to-day tasks.

While the situation is frustrating in itself, the unfortunate fact is that similar things are happening all over the open source scene. Frequently developers get too caught up in making sure that their code is "right" (that may mean designed correctly, stable, cool, standards compliant, well integrated, or anything else the developer feels is important), and not enough time is spent making sure that the product is usable. I suppose this is one of the draw backs to a development methodology where there is no external pressure to develop your product.

Usability is king, and trumps all other concerns in a product. If it's not usable, it's no good.

The data: URL scheme

Here's something you may not already know: You can include data directly in an (x)html page, instead of referencing it externally.

For example, most of the time when you want to display an image you would write code like this:

<img src="http://www.blogger.com/some_image.png" alt="some random image" />


Web browsers downloading your HTML source will download the text first, then download any external references, including "some_image.png" (assuming the user has not turned off image downloading).However, there are a few cases where you want to distribute an HTML file with images, but don't want to distribute multiple files. In those cases, the 'data:' URL scheme is what you need.

The scheme is documented in the (very readable) RFC2339. Essentially, you can include the binary data straight into your HTML code. The example they give in the RFC looks like this:


<img src="data:image/gif;base64,R0lGODdhMAAwAPAAAAAAAP///ywAAAAAMAAw%20%20%20AAAC8IyPqcvt3wCcDkiLc7C0qwyGHhSWpjQu5yqmCYsapyuvUUlvONmOZtfzgFz%20%20%20ByTB10QgxOR0TqBQejhRNzOfkVJ+5YiUqrXF5Y5lKh/DeuNcP5yLWGsEbtLiOSp%20%20%20a/TPg7JpJHxyendzWTBfX0cxOnKPjgBzi4diinWGdkF8kjdfnycQZXZeYGejmJl%20%20%20ZeGl9i2icVqaNVailT6F5iJ90m6mvuTS4OK05M0vDk0Q4XUtwvKOzrcd3iq9uis%20%20%20F81M1OIcR7lEewwcLp7tuNNkM3uNna3F2JQFo97Vriy/Xl4/f1cf5VWzXyym7PH%20%20%20hhx4dbgYKAAA7" alt="Larry" />


Which equates to this image:



There are many reasons why you wouldn't want to do this - it increases the size of your HTML file, forcing users to download more before they can see whether your content is what they want (especially if the embedded data is near the beginning of the file). There are also some limitations on the size of data and those limitations vary depending on where this technique is used. Still, it's a useful technique that can be used when you need to embed small amounts of binary data within an HTML file and you don't want to distribute multiple files.

piwup: A Picasaweb Image Uploader for Linux

One of my pet peeves has always been that unless you want to run google's picasa application under Linux, the only way to upload photos to your picasaweb account is via a klunky web interface that only allows you to select 5 images at a time. When I come back from a trip I have hundreds of photos, so this gets tiresome very quickly.

There is a kipi plugin that is supposed to be able to do this, but it has not yet hit the Linux distribution I am using, and I'm not about to start compiling plugins from source. Besides, half the fun is in making the application!

This is definitely not a finished application! I got it to the point where I could upload my images in a batch, but it needs more work before it's useful to anyone else. Here's a few sample screen shots:

Selecting images to upload.

Uploading the first image.

The application still has a long way to go. Just some of the things yet to complete are:

• Remove hard coded items from the code (account details, service host, album name), and make these configurable via a nice configuration dialog. Make sure password is stored in a secure form - via the KDE wallet perhaps.
• Make the GUI half-decent. Originally I just wanted something to work - I need to go back and do it again with a proper menu and image thumbnail support.
• Bug fixes too numerous to mention here... this is some rouch, cheap and nasty code!

Perhaps, once I get all this done I will attempt to get it officially released into some distros. I think it's a useful application, and the kipi plugin version doesn't seem to be moving along much. Yes, I realize that I'd be better off spending my time improving the kipi plugin, but to be honest I can't be bothered right now - this was a learning experiment for me as much as it was about making an application that solved one of my problems.

The entire application is written in C++ and Qt4. The more I use Qt the more I like it. This application was simplicity itself to make, and I look forward to continued development.

Why this website sucks: A rant on poor web design

I have just had an epiphany of biblical proportions:

Fixed-width websites suck.

Okay, so it's not a huge revelation, but still, I was quite proud of myself. Why is it that website designers think they know how large I want their website to appear on my screen? I have two 20" monitors, and many websites show content in less than half of my browser window.

Any decent website would have a template that showed content at whatever resolution the viewer wanted.


That's why there'll be some changes around here. I'm going to try and design my own blogger template, or at least rip of someone else's good work and call it my own.

Edit:

Round one of changes has been completed. The new theme is based on the stretch-denim blogspot theme, with a few revisions of my own.