Re: [Wikitech-l] Re: allow xml tags: a first approach

On Sat, 15 Jan 2005 12:58:42 +0100, Karl Eichwalder &lt;ke(a)gnu.franken.de&gt; wrote: WP's markup - or rather MediaWiki's - is a variant of the original WikiWikiWeb's, and was designed with ease of editting (by humans) at the very top of the agenda. There have been moves to create a standardised wiki markup - which would in no way resemble machine-friendly formats like XML, since the aim would still be human-friendliness - see, for example, http://www.usemod.com/cgi-bin/mb.pl?WikiMarkupStandard But for now, yes, it's "proprietary" to the extent that it is not developed in co-ordination with any outside body. As for "hackish", though, I'm not sure I agree: the *syntax* is perfectly good at doing what it does - which is to say, it allows people to describe the formatting they require of a piece of text. What's hackish is MediaWiki's *implementation* of that formatting - the parser which isn't a parser - being as it is based on a range of context-unaware text manipulations which continually interact in unexpected and undesired ways, violating users' assumptions about the syntax. The almost-developed flex/bison wiki-text parser (which outputs, being intended for computer use, XML) demonstrates that the syntax itself is amenable to far less hackish processing. Unfortunately, I have no idea what that example refers to. Are you meaning that wiki syntax could magically morph to suit a different language, with different formatting conventions, so people could copy-and-paste from one wiki to another? Or are you just saying that the markup should be more *semantic* (I don't see anything less "proper" about marking up what things look like; this is the way people view data), and the example is some debate about what formatting to use in a particular case? I think the key problem with semantic markup is that in order to have any advantage, it has to be both very rich and very standardised - e.g. you'd need to have a way of saying "this text is the name of a film", and people would need to know that this was "<movie>" and not "<film>" (or "<movietitle>" or "<filmtitle>", or any number of other possibilities). If you were to impose such a scheme on something like Wikipedia, you'd just create a tremendous learning curve. Obviously, semantic *representation* is good, and in some cases (such as Wiktionary, or the oft-mooted "Wikipeople"/"Wikibiography"/"WikiFamilyTree" ideas) I have my doubts as to whether a wiki really is the best way to go, simply because editting a block of text that describes its own semantic status is rather daunting for most people. Hence, as you say, XML editors - and more specifically, hence sites like IMDb that have subject-specific structure hard-coded into the software and the interface (if you edit a listing in IMDb, you never *see* the underlying representation, and the structure is defined for you so you don't need to). Again, you've lost me; for what purpose did they create this "box", in what sense is it "invisible", and how would XML/"a proper markup language" deal with the same issue? -- Rowan Collins BSc [IMSoP]