Daniel Veillard looks at XML from an Open Source perspective
20:07, 15 May 2003 UTC | Simon St.Laurent

While XML has benefited from free and open source tools since its beginning, the open source and free software community has taken a measured approach to XML. Daniel Veillard gave the XML-centric crowd at XML Europe 2003 a look at XML from a different perspective.

Rather than looking at the tools created by and for the XML community, Veillard stepped back and looked at what the broader open source and free software communities have done with XML. While some of this sounded like the familiar list of XML toolkits and markup-oriented processing, much of it focused instead on how different projects were using XML.

Veillard pointed out that while there are some systems built around the original notion of XML for Web publishing, most of the projects he discussed focused on other XML capabilities. While Veillard's employer (RedHat) includes over 3600 XML files in its Linux distribution, they serve a wide variety of purposes, largely for configuration.

Veillard noted that XML is becoming the default format for storing application information - after five years, developers now explain why they aren't using XML if they choose another format rather than why they are using it if they do. Tools and easier internationalization are crucial reasons for XML's popularity.

OpenOffice, KOffice, Abiword, and GNUmeric are all using XML-based formats for their files, often compressing it. Veillard noted that they all use different formats at present, but expressed hope for standardization and other interoperability approaches.

On the documentation front, most projects seem to have standardized on the DocBook vocabulary. While a WYSIWYG DocBook editor would be an enormous help, developers are still working with Emacs/PSGML and other tools - and building DocBook support into tools like Nautilus, ScrollKeeper, the KDE HelpCenter, and Yelp. KDE's xmlpo and GNOME's intltool help with translation systems.

While acknowledging Java toolkits, Veillard focused on the ready availability of toolkits like expat, his own libxml, QtXML, and the support built into Python and Perl. Developers have focused on the specifications that they need and can understand - Veillard emphasized the cost of sorting out complex or confusing specifications.

Schema validation has been a particularly difficult area, and Veillard expressed hope that RELAX NG might help simplify the process. XML catalogs have already been important for DocBook users, and XSLT and XSL-FO processing are also crucial for this group. XSLT is well implemented, while XSL-FO support is growing.

Veillard pointed out some surprising fields where XML had caught on, notably the use of SVG for scalable icons in both GNOME and KDE, with growing library and application support as a result. XML for user interface definition was also important, both in Mozilla's XUL and in GNOME's Glade interface builder. Unlike the commercial XML market, SOAP and XML-RPC web services haven't been taking off rapidly. Veillard mentioned growing Jabber support, however.

Veillard also explored some similarities between the values of open source and free software communities and those of the XML community. Openness and reuse are at the core of both approaches, and data and code reuse can be mutually reinforcing.

Veillard expressed some concerns about XML as well. A DocBook WYSIWYG editor is a major gap, and format standardization for applications is just getting started. Developers are also concerned about the dominance of Java and C# in the XML market, as much of their work relies on shared C or C++ libraries.

Veillard concluded by noting that XML's growing complexity may prove a barrier - "Open source developers are good engineers - they only do what they understand."

Re: Daniel Veillard looks at XML from an Open Source perspective (Daniel Veillard - 11:22, 20 May 2003)

Yes, OpenOffice is making progresses but the appoach is quite limited. Using XSLT to convert looses all entities definitions, and there is no image support. So it's not usable as-is for a lot of the existing DocBook documents in use. I doubt any solution based on XSLT to allow the editing process can really work. For editing native support and CSS based rendering seems quite a better approach.


Re: Daniel Veillard looks at XML from an Open Source perspective (Bart Locanthi - 15:40, 19 May 2003)

Regarding the use of compressed XML formats by OpenOffice et al..

I appreciate the sentiment, but this defeats the often-accompanying desire to have documents under source control.

Binary MS .doc formats are a lost cause, but XML, particularly if normalized in some way (no indenting?) is very diff-friendly.

Until CVS acquires knowledge of compressed text files as distinct from either text or binary files, programs that write XML files should decline to compress them, or at least offer the option to turn compression off.

Last time I looked, OpenOffice didn't allow this.

Re: Daniel Veillard looks at XML from an Open Source perspective (Andreas Kuckartz - 14:43, 19 May 2003)

Simon St.Laurent wrote: "A DocBook WYSIWYG editor is a major gap, ..."


DocBook Filters - Read and write docbook xml using

Re: Daniel Veillard looks at XML from an Open Source perspective (Conal - 01:48, 18 May 2003)

full URL (truncated in short comment above)

Re: Daniel Veillard looks at XML from an Open Source perspective (Daniel Veillard - 15:20, 16 May 2003)

Just to indicate that I have put the slides online at: they contains links to the various projects web pages.


P.S.: thanks Simon :-)

xmlhack: developer news from the XML community

Front page | Search | Find XML jobs

Related categories