Whatever the opposite of "I stopped reading when they said X" is, I did that when Nicolas drew a distinction between local and global semantics. Anyone who talks semantics without understanding that they require contexts to have any meaning at all is someone not worth listening to, and man oh man is the "semantic web" full of people ready and indeed anxious to make that fundamental mistake.
In the interests of counterbalancing the tendency to have nothing but criticism up in comments, let me just say: This is great stuff.
The topic here is completely unrelated to the semantic web, except in a tangential mention (the "etc"): "We can leverage the agreed “global” semantics of HTML elements, certain HTML attributes, Microdata, etc."
Your caricature of semantic web "people" is misplaced. Local specialized semantics is fully a part of the design. No one believes there is a single master schema.
I don't think parent meant the semantic web as in the Tim Berner's Lee version of it, he just refers those "semantic" html advocates (hence the quotes).
As for the semantic web in the TBL meaning, it is not worthy of caricature, anyway, considered the utter failure of RDF and XQuery outside of tiny niches, not to mention the flawed premises under it all.
Tiny niches? All the data tables in Wikipedia; Apple acquires Siri; Google acquires Freebase; the Google and Bing search indexes, .... This story is not over.
Bitter, absolute statements like that are regurgitations of a myth. (And maybe you mean SPARQL not XQuery.)
Which is to say, the Semantic Web didn't happen as promised, and it didn't happen in exactly the way you would guess based on my message, which is that where you had something that could create a local context it could happen in pieces, but no global whole has emerged, nor is there any particular evidence one is emerging.
http://reprog.wordpress.com/2010/09/02/bibliographic-data-pa... - read parts 2 and 3 as well. It may not seem directly related at first, but you should soon figure out why I consider this three-part blog series to be the thing that proves global semantic contexts are a hopeless pipe dream. (And remember that entire bit is about librarians; if they are that hopeless nobody else has a chance.)
Tiny niches? All the data tables in Wikipedia; Apple acquires Siri; Google acquires Freebase; the Google and Bing search indexes, .... This story is not over.
What those have to do with it? I mean, I don't know about the data tables in Wikipedia, but I'm pretty sure neither Siri, nor Google / Bing etc use RDF for their metadata and search indexes. It's been 15+ years now, the Semantic Web didn't caught on, and it has been shown to be misguided in the general scheme of things (could work for some specialized cases, where you control all the data, but even then, neither the tools nor the performance and market readiness are there).
Being able to cater for semantics != using RDF and TBL's "semantic web" notions.
And, yes, I mean SPARQL. Was working with a team back at uni that integrated XQuery with RDF/SPARQL and have them a little confused in my mind.
Sound advice, I found myself nodding at the various bits in the article, and recognize that over time I have gravitated towards similar patterns even in my limited amount of frontend work. The underlying idea is that CSS classes should be treated as rigorously as any other kind of identifier in a programming language: naming structure, precision in usage and clarity of scope are important.
In the interests of counterbalancing the tendency to have nothing but criticism up in comments, let me just say: This is great stuff.