|
as kendall correctly points out, the upcoming w3c workshop mixes two orthogonal issues, the question of a binary format, and the question of what xml really is.
there seems to be an increasing tendency to make the infoset the 'real xml'. i think that a proper information model would be a very smart thing to have, but i also think that the infoset is not the only way to go. and that maybe one should spend some time about making the infoset better (in particular, extensible, for example for being able to handle xml schema's psvi contributions)
for a long time, when people were asking about xml's 'information model', they were told that the bits on the wire were much more imporant than the model behind them, and that specs like the infoset were for spec writers only. as it increasingly turns out, if each and every new spec of the w3c is based on the infoset, then why not call this (which in essence is a mildly pre-processed subset of xml) the 'real xml' and 'xml 1.0' just a character-based syntax for it? this would make life much easier for many developers, who often think they are using xml, but in reality (through tools such as xslt and xquery) are using the infoset ("why can't i search all cdata sections, dammit!").
what i want to say (i got a bit carried away, i am afraid...) is that this workshop could be a good starting point to re-align some of the methods (and attitudes) of the past and get on with a proper and helpful separation of information model and representation.
|
Previous Message
