elliot's blog

del.icio.us - not so tasty?

When I first started using del.icio.us (a central, social bookmark repository), I was very impressed by the idea behind the service, and put a lot of effort into porting my bookmarks into it (even writing some scripts to help me automate it). However, I've not used it that much recently. Why? Because I've had several occasions when the service has just disappeared, leaving me with 500 Server Errors, or 503 Service Unavailable.

This has made me realise that for something as important as bookmarks, there is no room for down time. I use them without thinking when they are part of my browser, and know that when my browser is working, so are my bookmarks. When you put your bookmarks onto an external service, that service needs to be as reliable as a web browser (i.e. work every time without fail). As soon as the black hole of system unavailability looms, however small, the system becomes unusable.

I've briefly considered writing my own clone of del.icio.us, which people can install themselves and use from their own web space. This would obviously remove the social dimension; but, to be honest, I don't care about this anyway (I'm not going to browse around other people's bookmarks, as this seems tantamount to obsessive compulsive disorder). All I need is a shared bookmark service I can access from any of the multiple internet devices I use (5 at the last count). This is one of the chief failings of del.icio.us: it's not open source, so I am powerless to manage a service of my own, where I can control the availability and quality of service.

IT Conversations

If you've not come across the IT Conversations site before, and you work in IT, it's well worth a look. The site is free to use, and contains downloadable (mp3) and streaming versions of interviews with IT luminaries, plus recordings of presentations from conferences.

My personal favourite so far has been the interview with Bruce Schneier, security guru. Full of interesting analysis of 9/11, plus some particularly good "factoids" about risk perception (e.g. more people are killed each year by pigs than by sharks!).

Google Scholar

In a previous incarnation, I was a PhD student at the University of Birmingham (my thesis is still available from the Computer Science Department website). Why do I mention this? Because today I heard about the Google Scholar service, which is currently in beta. And, like all vain academics would, I looked up my own publications using its search facility.

I was immediately impressed: somehow (not sure how) only research papers I had written were returned as search results. Usually if you put my name into a search engine, you get endless Elliott Smith (the singer-songwriter who apparently committed suicide - note the different spelling) websites. In addition, Google had also managed to connect my results to a paper by another researcher which referenced them. It was interesting to see someone else's review of my work, which I'd never come across before. I disagreed with it, of course... If only it had been around when I was doing my PhD!

BloGTK

The web interface in Wordpress is all well and good, but if you want a desktop application for posting to your blog, BloGTK might be the thing for you.

It gives you a nice word-processor-like interface (written in Python/Gtk), which provides all the functionality of the web interface, but is significantly easier to use. The preview function is nice as well.

I think it only runs on Linux, mind.

A couple of things to mention:

  • The server URL should point at the Blogger API URL for Wordpress; this will be http://your.domain/path/xmlrpc.php. This caught me out the first time.
  • Select Movable Type as the Blogging System. If you select Blogger, it doesn't work as well.

Open source CMSs

I am very confused: there are so many open source CMSs, and they vary so much in quality, features, appearance, complexity, and across multiple other dimensions. Finding one which does everything well is extremely tricky. I've been reviewing many of these beasts over the last couple of weeks, and have installed about 10 and tried to build the same simple site with each (using my "20 point CMS test", which I may publicise eventually). My eventual aim is to produce a report recommending a selection of half a dozen, for a variety of different needs.

So what can I recommend so far? Well, my current "generic" leader is Drupal. This is well-thought of in the community; the code seems relatively clean and fairly easy to extend; it has a lot of nice modules for doing all the common CMS tasks; the default templates produce decent XHTML/CSS. What's missing? WYSIWYG editing in Mozilla/Firefox seems quite an important omission. The blog module is also a bit primitive by comparison with WordPress (which I really like), so I think I'll stick with WordPress for blogging at the moment.

My next self-imposed challenge, then: see how hard it is to write a Drupal module which replaces Drupal text areas with the TinyMCE in-browser, WYSIWYG HTML editor. I've seen this done in Xaraya, but not for Drupal yet. This could actually be a useful module for other people, and give me a concrete project which will encourage me to learn how to code Drupal modules.

Website launch

I've recently been building a website for a charity in Manchester (see Self Help Services), which I launched last week. I found out about the project through the IT4Communities website, which I recommend for anyone with IT skills who would like to do some charity work. I also got involved to improve my portfolio, when I was thinking about a career change and considering consultancy.

One thing I'm quite pleased about is that I designed it to the W3C Web Content Accessibility Guidelines. As a consequence, when I tried the site through my new HP iPAQ PDA, it displayed very nicely, unlike 90% of sites on the web. I've always maintained that building accessible sites would future-proof them: human-accessible = machine-accessible as a consequence. Nice to see it working in practice.

Spread Firefox campaign

Made my donation (worked out at £17) to this campaign: Spread Firefox - Igniting the web.

My name will appear, along with around 9000 other people's, in a New York Times ad. in support of Firefox. If you haven't already switched, why not?

SugarCRM bug report

Finally, I've made a concrete contribution to an open source project: I submitted a bug report to the SugarCRM people a couple of weeks ago, and it's now been fixed. SugarCRM is, as you might have guessed, a CRM system. It's interesting because, while it is open source, it has a huge amount of backing from venture capitalists. It's something I intend to review before too long, as I've been given the task of writing a CRM "primer" as part of my work schedule. I'll be looking at it alongside XRMS and other projects.

Anyway, it's nice to finally make a contribution to open source, even though it's small.

(The bug report is at SourceForge.net: Modify:1031310 - SQL error when associating contact with task.)

del.icio.us extensions

Found an interesting Java API for working with del.icio.us at: http://delicious-java.sourceforge.net/.

I've also been working on some Python scripts for parsing a Mozilla bookmarks file and posting the results to the del.icio.us REST API. Take a look at the scripts if you're interested.

Then found someone else had been hacking around with Python and del.icio.us: http://www.hackdiary.com/archives/000060.html. Should have searched for that before I started, I suppose. I just got carried away. Still, I haven't found any Mozilla to del.icio.us parsers other than mine.

del.icio.us

del.icio.us is a bookmarking service (see http://del.icio.us/doc/about). Basically, you can put your bookmarks into a central web-based repository, and use them from any browser (e.g. by opening your bookmarks in an HTML page). There's a link to my bookmarks in the menu on the right.

Plus points for me are:

  • Firefox/Mozilla integration: you can get yourself an extension at http://mozdev.org/ which adds del.icio.us functions to the context menu; for example, the "Post current page" option posts the current link to the bookmarks server.
  • REST API: you can programmatically access your bookmarks as XML feeds, with the option of using parameterised queries.
Syndicate content