On Fri, Apr 30, 2010 at 04:17, Jakob Voss <jakob.v...@gbv.de> wrote: > But all the flaws of XML can be traced back to SGML which is why we now use > JSON despite all of its limitations.
Hmm, this is wrong on so many levels. First, SGML was pretty darn good for its *purpose*, but it was a geeks dream and pretty scary for anyone who hacked at it not fully getting it (like most normal developers). As with many things where the learning curve is steep, it fell into the "not good for normal consumption" category and they (well, people who cared, and made decisions about the web) were "forced" to make XML. But JSON? Are you sure you've got this figured out? JSON as a object serializing format is good for a number of things (small footprint, embedded type, etc.), but sucks for most information management tasks. However, I'd like to add here that I happen to love XML, even from an integration perspective, but maybe that stems from understanding all those tedious bits no one really cares about about it, like id(s) and refid(s) (and all the indexing goodness that comes from it), canonical datasets, character sets and Unicode, all that schema craziness (including Schematron and RelaxNG), XPath and XQuery (and all the sub-standards), XSLT and so on. I love it all, and not because of the generic simplicity itself (simple in the default mode of operation, I might add), but because of a) modeling advantages, b) cross-environment language and schema support, and c) ease of creation. (I don't like how easy well-formedness breaks, though. That sucks) But I mention all this for a specific reason ; MARCXML is the work of the devil! There's a certain dedication needed for "doing it right", by paying attention in XML class, and play well with your playmates. This is how you build a community and understanding around standards; the standards themselves are not enough. The library world did nothing of the kind ; http://shelter.nu/blog/2008/09/marcxml-beast-of-burden.html The flaws of XML can most likely be traced back to people not playing well with playmates, and not the format itself. > May brother Ted Nelson enlighten all of > us - he not only hates XML [1] and similar formats but also proposed an > alternative way to structure information even before the invention of > hierarchical file systems and operating systems [2]. Bah. For someone who don't see the SGML -> XML -> HTML transgression as an inherited and more rigid structure (or, by popular language, more schematic) as a document model as a good thing, I'm not impressed. Any implied structure can be criticized, including pretty much any corner of Xanadu as well. (I mean, seriously; taking hypermedia one step closer to a file system does *not* solve problems with the paper-based document model of HTTP, it just shifts the focus) > In his vision of Xanadu > every piece of published information had a unique ID that was reused > everytimes the publication was referenced - which would solve our problem. *Having* an identifier doesn't mean that identifier is a *good* one, nor that it solves your problem. There's plenty of systems out there where everything has an identifier (and, if you knew XML deeper, you'll find identification models as well in there, but people don't use them because the early onset of XML didn't understand nor need them). Have a look at the failed XLink brooha for something that worked and filled the niche, but people didn't get nor did tool-makers see the point of implementation, and the thing died a premature death. The current model of document structure and XQuery is somewhat of an alternative, but people are also switching to CSS3 styles as well. The thing is, just because you've got persistence in a system of identifiers, it does not follow that the information is persisted; the problem of change is *not* solved in neither systems, and so we work with the one we got and make the best of it. One thing I always found intriguing about librarians were their commitment to persistent URIs for information resources, and use of 303 if need be (although I see this mindset dwindling). I think you're the only ones in the entire world who gives a monkeys bottom about these issues, as the rest of the world simply use Google as a resolver. I can see where this is going. :) Regards, Alex -- Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps --- http://shelter.nu/blog/ ---------------------------------------------- ------------------ http://www.google.com/profiles/alexander.johannesen ---