Just composing proper XHTML was never enough to be very useful. The way that XHTML was promoted to web developers (my impression, I was only around for the tail end), was as merely a stricter version of HTML, and that's how IE implemented it, so that was the only practical way of using it.
XHTML on its own does hardly anything that HTML didn't do. It's easier to parse, but HTML was already parseable and anyone who was going to try to extract meaning from human-readable webpages could already do that. The real value of XHTML was that it could be generated from domain-specific XML using XSLT. So your data could be served as machine readable XML in a standard, versioned schema particular to the application or document publisher, and then converted into an XHTML webpage by the browser. IE didn't implement XSLT for too long, so nobody did this, but the idea behind it is a popular approach today. It's essentially the same as serving static HTML which then fetches its data by API queries using javascript. But this would have worked with javascript disabled, and would have made the human- and machine-readable data exist at the same URI, so that it would be clear to machines what data the human-readable presentation was representing.
Ideally content providers would also serve an XSLT transformation for producing RDF/XML from the domain-specific XML representation, so that generic web crawlers could understand the meaning of the data on the page, or the generated XHTML could contain RDFa markup. We still havent gotten to the point of re-engineering this functionality, and it may never happen, since the major web businesses' revenue model depends on maintaining exclusive control of their data, and the interoperability ideals of linked data are directly opposed to that.
> The real value of XHTML was that it could be generated from domain-specific XML using XSLT. So your data could be served as machine readable XML in a standard, versioned schema particular to the application or document publisher, and then converted into an XHTML webpage by the browser.
Ok, now I understand all the buzz around XHTML at the time I was a student, and why so many people kept talking about XSLT. I just never saw anybody pointing it before, and for something that interesting I have no idea how come people weren't talking about it everywhere.
>HTML was already parseable and anyone who was going to try to extract meaning from human-readable webpages could already do that.
Extracting data from circa-2005 HTML was a nightmare. It's only better now because libraries like beautifulsoup have gotten so much better at guessing structure, and even today I have things that just plain come out wrong because the HTML structure of what I'm scraping is so bad.
While I agree those were the goals, you don't need XHTML and XML for achieving these. The same is possible with SGML and/or other processors (and xslt can at least generate HTML, too). HTML and RDFa crawling on the web is a thing.
Saying as someone who coded a web app with browser-side XSLT and linked data ten years ago.
XHTML on its own does hardly anything that HTML didn't do. It's easier to parse, but HTML was already parseable and anyone who was going to try to extract meaning from human-readable webpages could already do that. The real value of XHTML was that it could be generated from domain-specific XML using XSLT. So your data could be served as machine readable XML in a standard, versioned schema particular to the application or document publisher, and then converted into an XHTML webpage by the browser. IE didn't implement XSLT for too long, so nobody did this, but the idea behind it is a popular approach today. It's essentially the same as serving static HTML which then fetches its data by API queries using javascript. But this would have worked with javascript disabled, and would have made the human- and machine-readable data exist at the same URI, so that it would be clear to machines what data the human-readable presentation was representing.
Ideally content providers would also serve an XSLT transformation for producing RDF/XML from the domain-specific XML representation, so that generic web crawlers could understand the meaning of the data on the page, or the generated XHTML could contain RDFa markup. We still havent gotten to the point of re-engineering this functionality, and it may never happen, since the major web businesses' revenue model depends on maintaining exclusive control of their data, and the interoperability ideals of linked data are directly opposed to that.