HERE are a variety of technique for summarising page content. Excerpts may be considered one of them, metadata in the (X)HTML header might be another. There is also a sharp rise in the use of tags, which can easily infer the ‘theme’ of a page a and can cohesively reflect on trends across sites (confer tag clouds or see image below).

I am discovering more and more services that are beginning to rely on a succinct collection of keyword, much like tags in Technorati, del.icio.us or the new meta search service gada.be. To each page, a concise representation simply gets bound. Prepare for more of that tagging phenomenon to be seen in the future. In its absence, pages become less desirable as they are more bandwidth-consuming.

Contextual tags cloud in July 2005

Finally, and perhaps more interestingly, advertisements in a page can be made more relevant by using tags, having manually embedded them in the page. This avoids advertisement from appearing where they would become a contextual misfit. Thus far, however, I have only come across support for tag-guidedads in WordPress. As tags are often generated automatically, e.g. derived from the page using scripts/tools, I can envision the same ideas being extended and exposed to the entire World Wide Web. Google AdSense makes an attempt at finding out for itself what a page is primarily about. It does so off-line or ‘on the fly’. Why not involve the user and use his/her knowledge for assistance? That is where tagging, as in the example above, bears tremendous potential.

PDF‘s, text files, HTML‘s and
directories in the KDE Desktop with previews (click to enlarge)

ERTAIN issues arise when habits and user orientation in his/her desktop are interfered with. Desktop environments, installers, filesystem structures, or even platforms in general are often more workable and thus successful if they comply with the expectation of new users. What if these are made too stringent by the developers, however? What if decisions and conventions are voted for without involvement of the end-user?

Many desktop environments, free ones in particular, are not made uniform. The resulting diversity and flexibility leads to difficulties in using somebody else’s settings, i.e. working in conjunction under the same session. Some GNU/Linux distributions include more or less any desktop environment which exists while settings are kept apart and well-separable for each user of the system and each desktop environment. So, where does the the problem lie? Give users more freedom, I suggest, and make that the norm. It is, after all, their own computer and they understand their needs better than the developers of the operating system.

What appears worse than all is the scenario where decisions are arrogantly by the vendor and then enforced. This is often the path that operating systems such as Mac O/S (to its varieties) and Windows do. Give the user some more choice, I say, by adding options for endless customisation that has very few boundaries, if any (Open Source). Give all users the freedom they deserve and allow them to express individuality and adapt/tailor their desktop environment to suit their needs. Menu entries and other widgets will remain unaffected, so documentation, for instance, will not suffer as a consequence.

People work differently on a variety of applications, for a wide variety of purposes. The domains in which they work differ as well. A person working with many open windows at any given time might prefer to have “focus follows mouse cursor”. Contrariwise, to some, this “focus follows mouse cursor” behaviour is highly adverse to habits. To list yet another example, a Web server needs to have a light desktop environment that is less susceptible to breakage and consumes small amounts of RAM.

These considerations are all highly defensible. If Linux was to be deployed in more public clusters, for example, choice could (and should) be given as to which desktop environment should be used. Different strokes for different folks, but all can be catered for provided that disk space is made available. The pertinent settings would reside in the home directory of each user. Thumbs up to Gnome, KDE, and the rest of the self-motivated window manager teams around the world.

Related item: KDE Versus GNOME

A long time ago I mentioned top software design/usability bugs. It occurred to me yesterday that there is yet another common deficiency, which is the inability to undo selection and deselection of files, entries and the like. This should become a very fundamental functionality in my opinion.

How many times in the past did you use the CTRL or SHIFT keys to establish and highlight a collection of files? One wrong click and the entire selection is gone, leading to a habit whereby files are handled in subsets, i.e. in smaller batches. In all operating systems I have come across, a selection of files, for example, forms a newline-separated list of the files with their full path. All of this is stored in the clipboard, so implementation of an undo stack should be trivial and incur no efficiency penalty.

ARLY this morning I bought iuron.com. The host was admirably rapid in its automated response and setup. Subsequently, after just minutes since the purchase I was able to access the domain (DNS was up-to-date already) and initialise things, e.g. content, passwords, customised error pages, siteinfo.xml, robots.txt, favicon.ico, etc.

In the intermin I created a site/project logo using The GIMP. It took me roughly 10 minutes to produce the compact GIF version, of which I keep a non-lossy version (bitmap) as often required.

A detailed but not yet comprehensive proposal is bound to go public in the brand-new domain soon. The specifications are not public and are password-protected for the time being. Let the title and excerpt give you a rough indication of where the project is headed. I am yet to seek funding opportunities so many more details are (hopefully) soon to follow.

Many on-line services incorporate RSS feeds these days. Briefly observe, for instance, the vendor-specific feeds management for this Web log.

It turns out that Google have joined that game with what they call Google Reader. GReader is bound to deliver quite a lot given what we know about Google engineers and their love-affair with Web applications (more latterly videos).

For feeds, I am still using RSSOwl (JRE, hence inter-operable), of which I am a tester. I am also a tester of the Web-based, AJAX-rich Feedlounge (screenshot below), but admittedly I never contribute immensely as a tester. I use RSSOwl (screenshot below) almost exclusively. I am rather loyal to RSSOwl though I occasionally take Feedlounge for a short spin. Feedlounge is somewhat irresponsive, is still in its testing phases, facing an uncertain future and boasts commercial aspirations.

Feedlounge in action

RSSOwl back in April (click to enlarge)

EVIATING from the a development branch, thus staying away from the regular updates cycle, is a step which is bound to have its problems, e.g. security and missed out extensions. I am fully aware of these issues and I made mistakes in the past nonetheless, particularly with Web applications. I am rather confident and comfortable with my choices to extend applications, but one must remember the pitfalls, as well as the very few advantages.

Rather than simply downloading pre-installed and/or pre-configured packages like Gallery, WordPress, a VPN client or PHP-Nuke, one can add a wee bit of ‘individuality’ to them. For example, with actual words and vocabularies that do not suit your way of speaking, you can instill some personality within the software — a beauty to be found most commonly in Open Source software. It gives a feeling of ownership, a sense of personal contribution.

Linux users are often inclined to hack software and customise it, maybe because they can. With some Linux distributions like Gentoo, people struggle to compile individual components and some have the odd habit of re-compiling the Linux kernel. Again, it gives an utterly unnecessary sense of achievement. That free platform, however, unlike other platforms, can be claimed to be one where very little intervention should be involved. With 5-CD installations, for instance, all the necessary software is in place, is well-integrated and simply works without any apparent conflicts. Somebody else took care of all the nasty installations and then burned some ISO to be distributed and put as-is on other people’s computers. So, we perceive a situation where there is a great deal of freedom, but at the same time that freedom is unneeded or even discouraged.

Change to existing packages is risky. It can bring an application to a fragile state. Attempts to exploit vulnerable (usually non-existent if the owner is fortunate) components are being re-directed to an unpleasant page at server level on this domain. I hope these can successfully deter a determined hacker. The only ones to come across such a page would be those attempting to sabotage (hence the rel="nofollow" in the link above).

When forking and choosing to no longer follow a development cycle, there is always some possibility of ‘flushing’ self-made components and starting fresh with similar, even identical data and an application belonging to the current generation, hence incorporating state-of-the-art technologies. Upgrades usually mean some loss of personal changes which are better off never lost. This may lead to feelings of regret due to waste. It also means hours of conversions, not to mention the laborious process that is involved in updates and getting accustomed to new interfaces. As it stands, I continue to fork, but do so cautiously.

When I first set up this site, which is administered using cPanel, I was blocked access to a terminal (AKA shell, console, command-line) and denied telnet/SSH access by the Web host. By default, most hosts do likewise, depriving site owners from excessive flexibility in hope that security will not be jeopardised and sites will rarely ‘break’. Nonethess, I found a terminal workaround, which I would like to share.

Run command as a cron job (widget/icon shown above). Have it scheduled to run a minute later with output to get sent by E-mail (default option) or saved to a file, e.g.

du -all >~/output_file

output_file, which gets created in your home directory, should then contain text that otherwise will have appeared in a terminal. Likewise, virtually anything including installations can be carried out. System administrators would become distressed if this practice became commonplace. In essence, this is the same as having a terminal-based interface, albeit it is slower and unresponsive.

Example cron job in cPanel – periodic mySQL backups (click image for full-sized equivalent)