Xtech 2005, Little bangs, Rip mix & share, Just do it

Rip, mix and share is the quote I took with me from her presentation.

I appreciate discussion on licensing and copyright in this conference. There's a lot of copyrighted material owned by governments (indirectly) that is there for the benefit of the public. It makes sense to allow people to use that material in various ways. This does not only apply to television history at the BBC, but also to art in government funded (national) galleries, for example.

The various Creative Commons licenses enable this process, within a proper legal setting.

little bangs is the Mozilla strategy. Bring out features at a time. Advance standards one step at a time.

I visited Rob's presentation to get a feel for the kind of market Avalon is aiming for. I'm still not sure if this technology will have anything to do with websites or webpages ever. In some ways it competes with CSS/SVG/Flash, so yes we might be seeing XAML objects embedded in webpages. And it might replace the horrible HTML/javascript sample code that MSFT now provides for Media Center. But mostly, it seems to be an XML layer for conventional application development, somewhat analogous to Glade.

Some quotes: from a 2 foot to a 10 foot UI experience, new platform [...] to take advantage of GPU, the website is no way near as impactful as tv

The markup models the UI hierarchy, presentation timeline, etcetera. In that way, it's similar to SVG and SMIL. Various elements in the XAML tree take advantage of the functionality provided by the underlying .NET components. There are no WYSIWYG design applications for Avalon yet, MSFT expects that there will a lot of tools available for Avalon. As of yet, I haven't seen a MacroMedia Flash equivalent for Avalon. Avalon will be released in 2006.

The XAML markup for Avalon is not semantic, in my opinion. The Media Mania demo would be completely different if it were done in current HTML/CSS; The HTML would be a semantic rich document, and the CSS would hold all the eye-candy instructions. This makes XAML stuff (a) hard to scale down a XAML document for "lesser" devices and (b) hard to sell for an exceedingly critical audience hammering on accessibility. From what I understand, MSFT is not aiming for (a).

I don't know how I should mix or convert an HTML/CSS site with XAML. I am not even sure if XAML or Avalon have anything to do with websites. Most television-like concepts are built on a combination of HTTP and HTML, but I'm wondering if there's a place for the World Wide- or even Semantic Web in these concepts: they seem enclosed within the worlds of broadcasters and cable operators.

So it seems to me, Avalon technology will not be important to me, and can safely be ignored for a while.

Unfortunately Nick didn't get his laptop to work with the beamer. Nick held a talk about the various Apache modules he has created. One of the interesting topics was content filtering with SAX, driven by namespaces. SAX is used in mod_proxy_html, for example.

Nick's reason for not using a scripting language based glue layer, is performance, he said. More importantly, I think, his modules fit in with the filter concept of Apache 2.0.

Apache's filter concept can be compared with Cocoon, but it's application is much more diverse since it operates on request and response message in the HTTP sense, and not on XML messages [my opinion].

Håkon made the point of being able to use a webpage on a variety of devices. CSS provides a way for this without changing the document.

This is a very powerful concept, much more than the idea of transforming documents on the server to suit the client. That is error-prone and always results in stupid work for the programmer, usually resulting in one good document and several lesser descendants.

With CSS, the client application can decide how it is best able to render the document.

This has become more relevant today than ever before, with the plethora of devices coming out that do web browsing of some kind. Just look at the latest devices like the Nokia 770 Internet Tablet. All these devices have varying form factors.

I had never heard of Access. Access actually does a lot of mobile phone handsets in the east. Access's GUI is completely based on browser technology (HTML, CSS, javascript), even parts like the handset start menu. This makes it easy for mobile operators (i.e. telephone companies) to change the user interface to their liking.

Axel held a talk about the RDF engine implementation in Mozilla. Axel works on the redesign of this implementation. His new API's should lower the entry barrier for people who want or need to work with Mozilla RDF (basically anyone who works on Mozilla or a Mozilla extension).

For more information look at the RDF:Home Page on the Mozilla Wiki.

E4X stands for ECMAScript for XML. It is an extension to javascript (ECMAScript). Basically it allows you to write literal XML inside javascript code blocks and it has an obscure syntax for doing DOM and XPath work on the XML document. To put it friendly: this syntax is as concise as PERL. Or, as Brendan put it: There's a lot of magic built into it. Just look at the examples in his presentation.

E4X will be part of Flash player 8, Jscript.NET, and Firefox 1.1 up. E4X is an ECMA standard, ECMA-357.

I guess Flash programmers/designers will love the XPath alternative in E4X.

Apple's Safari webbrowser and upcoming Mozilla Firefox releases have builtin support for Scalable Vector Graphics (SVG) and the canvas element (<canvas>). Opera is also working on support for both technologies. The ForeignObject technology makes it possible to re-include HTML inside the SVG or Canvas components.

This means that web designers can combine various graphics and animation techniques inside a webpage. This has many advantages!

Content no longer has to be offloaded into binary bytecode for Flash movies. You can just mix content and graphics instructions inside the single webpage. When clients don't have the necessary support for the graphics, e.g. text browsers, it's easy to degrade gracefully the page with the same content.

It's far easier to position graphics in a mixed webpage, there's no need to reserve square areas in the webpage for Flash movies; You just use existing CSS possibilities for positioning and scaling the graphics.

This is possible by using mostly existing W3C standards, so you instantly get W3C goodness.

This technology is actively being used in the Apple Mac OS X Dashboard widgets, for example.

Together with javascript and XML, SVG and canvas will become key parts in web application development and confirms the web once more as the dominant platform for application development of this age [my opinion].

The two technologies, SVG and canvas, seem to compete. Both aim at rendering graphics. The underlying concepts are quite different however. Hence most people should have no trouble choosing which technology is best for a certain project. SVG uses a DOM model, making it easy to store and distribute the graphics, but quite difficult to include a large document in a webpage. Canvas is a programming interface to the window graphics. This is convenient for rendering dynamic data such as clocks and stock ticker tapes, and it can be used to programmatically duplicate, remove, add, or alter (SVG) graphics.

Is XHTML2 feature complete? Well, I guess so. It has various changes in the meta data model over HTML4, making it better suited for the Semantic Web than HTML 4. It takes over many of the existing HTML features.

It is not a radically changed specification. There still are the <head> and <body> elements, even though many people are unsure what their purpose is.

The spec seems well received by the audience. With this presentation, XHTML2 went Working Draft.

My assumption is that many organisation will want to migrate to XHTML 2 quickly, in their editors and content management backends. It's support for namespaces, metadata and so on is so much better that it becomes a much better source document for many purposes.

Somewhat opposed to the previous presentation, Ian introduced WHATWG as a platform for browser makers and interested individuals that work on a hypothetical HTML5.

WHATWG aims at standardising present hacks, additions and acute shortcomings in HTML 4, specifically for web application development.

WHATWG suggests a number of new element names, such as <datagrid>, <canvas>, <event-source>.

It is not entirely clear what WHATWG's position is on XHTML (2). Since the suggested elements are not and will not be in XHTML (2), the audience was amused or abhorred by Ian's presentation, depending who you asked.

Beagle is somewhat like Apple's Spotlight, in that it is a search engine for your personal computer.

Jon's notebook crashed the evening before this presentation, so he couldn't demo Beagle. You can find some Beagle demo's on Nat Friedman's website.

Jon is a great guy, presenting potentially boring programming stuff with flair.

Of course, Jon did the obligatory Mono / C# plug. Beagle uses the Lucene.NET version of Lucene for the indexing work.

Initially, some years ago, the BBC offered a javascript nugget that you could add to your website to present BBC news on you own website. This would retrieve the news from the BBC website, and render it in a manner that the BBC approves of. According to the license that came with the nugget, you were not allowed to change anything in that nugget.

After the BBC News Interactive team added an RSS newsfeed to the site, slowly people began to use that feed as source for the BBC news nuggets on their websites. But since that feed does not include any styling instructions, the BBC was afraid that the BBC brand might not be properly represented on those websites.

It turned out, however, that people were actually doing very nice things with the feed. So, now the BBC stance is Built what you want using BBC content.

Kevin showed some stats from their sites; The BBC News Interactive service gets about 16 million clickthroughs a month from the RSS feeds. That is 3% of visitors. Since the Firefox release with Live Bookmarks, the number of downloads of feeds has exploded.

Here's an example of one of their feeds, nicely layed-out with CSS: BBC Sport UK Edition XML Feed

I'm not really into RDF, but I always have this suspicion that I should be interested.

This (vague) interest is mainly fuelled by research work I did about ten years ago. I was looking into computer support for conceptual design and collaborative work. I found that there were many (computer) models of design work that were too limiting for actual practical use. Ontology, the Big Word coined by Aristotle, seemed like a way out for research groups that are forever modelling some practises they see out in the wild, only to find out that the resulting applications would only cover 20% of the work of 20% of practising designers. The other 80% actually hate you for not understanding their work and seemingly trying to make them work with your Orwellian application.

At that time RDF was in its infancy. I think I read some information about it, but wasn't able to see if this model was enough formalised to be able to do interesting stuff with it.

Back to the presentations. David Wood presented a database system capable of storing RDF notations. Given that I don't know much more on this subject than the triple subject-predicate-object concept, this talk was way over my head.

The other David (Beckett) presented his libraries for working with RDF triples. His starting point is great, IMO, writing C libraries with bindings to various other languages and making this available with a free license. First of all, if I were to experiment with RDF I really would mind putting a lot of data into a datastore I don't know yet. Secondly, I would want to experiment in a language that makes it easy to experiment, e.g. Python, Perl or the command line. I did miss a real world application of his work.

For now, I think I don't have any application for meta data stuff at my work as web developer. I find it an interesting topic however, and I certainly am going to search for some cases that might prove that I should be using it.

Stacking XML processing steps, such as but not limited to XSLT, in a pipeline improves the manageability of XML publishing. Good story, pretty much what we do in our work. Jeni emphasised that there are a number of products available to work with XML pipelines.

I can only add that I wrote this super-simple, Makefile-driven publication pipeline hack (reusing existing tools, very much in the UNIX spirit of doing things).

Jean summarised the topics and general feeling of the conference nicely. Jean added two important statements:

Say NO to binary XML. (Followed by applause from the audience!)

People like to think in terms of documents! Which is something to think about for all those web application developers out there... if they agree with Jean.

He also went on too long about the merits of working with XML in Office 2003.

My comments above are only a small part of what I took with me and absorbed. I'm still struggling to create one holistic opinion about the whole conference... It was a pretty overwhelming experience.

I think the conference is expensive at €1300,-, but it is also packed with interesting presentations.

One of the themes I think I recognised was that a lot of technology is in place to create a semantic web and to take the web in general to new places, like the mobile space.

There are a lot of competing technologies and standards. I guess this competition is good, and we'll see which standards, applications etcetera will survive in a few years time. Given that these are mostly open standards, there isn't much fear for vendor lock-in. Often, there is some sort of interoperability or conversion method in place. Hence the just do it mentality that was in the atmosphere.

The little bangs theme echo-ed through in many presentations. Incremental development seems like a natural thing to do with so many good open standards and open source software libraries in place. Step by step improving and integrating those components in a modular fashion makes it manageable to create innovative platforms that built upon an already impressive body of work, called the world wide web.

Industry presentations are cool. The BBC talks by Paula, Matt, Ian and Kevin were very much appreciated. Such talks give insight into the adaptation of new technology in an organisation.

I would have like to have seen the presentation by Dominique Hazaël-Massieux titled "Bridging XHTML, XML and RDF with GRDDL" on Gleaning Resource Descriptions from Dialects of Languages (GRDDL), which was at the same time as Ian Hickson's presentation.

Matt Biddulph also seems to have presented a nice presentation on the directions of the World Wide Web and what the BBC is doing in that respect; presentation here: The Application of Weblike Design to Data, Matt Biddulph

What else would I have liked to have seen? Well, given that the conference was already so densely packed with information, I wouldn't want to see more presentations. I would love to see a similar event for more protocol oriented topics like HTTP and its friends webdav and caldav. Or perhaps even lower level stuff like STCP messaging. Perhaps I should consider visiting ApacheCon, since that would seem like a nice addition to xtech for a web developer like myself. Real web junkies might also want to visit reboot 7.0 or @media 2005 on Web Standards and Accessibility. Or for the more scientifically correct visitors: European Semantic Web Conference 2005. It's festival^H^H^H^H conference season in Europe!