I agree with Matt.
The thing is, building your own (proprietary) XML architecture may sound tempting at first. You can create all the element names so nicely yourself, and they'll be called what paragraph formats used to be called.
I started the same way a long time ago. But it quickly gets out of hand, especially when there are many authors ("But I really, really need this additional element," "Nah, but this element must be allowed in this place too!").
After a few years, you have a colossal architecture with hundreds of elements, a huge DTD/EDD, lousy performance, authors who can't see through all the elements. Somehow it is still XML, but there is no real structure anymore. And there are more and more problems: updating stylesheets for authoring and output, dragging element definitions in the filters of translation tools, adapting more and more complex XSLTs, e.g., for SmartPaste, and so on.
At some point, no one can keep track anymore, and in the worst case, the original architect doesn't even work in the company anymore. Specialized developers are then brought in for a lot of money to clean up and simplify the whole thing so that the authors can work halfway productively again. And in the end, everyone ends up with DITA.
(That's about the 10-year curve I've seen in I don't know how many companies).
Over the last 10, 15 years, I've become more and more of an opponent of "custom XML" and more of a fan of standards, especially DITA. The advantage is that an international standard like DITA not only has broad tool support (starting from CCMS systems all the way into the translation supply chain), but it's easy to find authors who feel at home in it. There are enough specialists when particular requirements are needed.
The international OASIS committee maintains DITA. Following the detailed discussions in the committee about specific elements and attributes, it is easy to understand the intellectual effort that goes (and must go) into developing an XML architecture that can be used as widely as possible. I think it is less and less a good idea – from an economic as well as a technical point of view – to want to make this effort in one's own company for one's own architecture.
Yes, you may have to compromise on a standard architecture, and reduce the (usually only supposedly necessary) complexity, and break down the hundreds of individual rules to a few. But it is worth it!
... View more
AdChoices