XML, Java, and the Future of the Web (1997)

Looking back at XHTML I am reminded again and again that the best technical solution often loses out.<p>Instead we have HTML5 which contains nothing that couldn&#x27;t have been easily expressed in XHTML but have none of it&#x27;s strictness that would have enabled much simpler and better browser implementations.<p>The arguments against XHTML amounted to &quot;but wahh it should still work even if the syntax is wrong&quot;. Thing is none of the major proponents of XHTML were advocating for exterminating the existing doctypes and their associated quirks modes. People could continue to use that but XHTML would be there, ready for the advent of non-shit programming to hit the web.<p>People are starting to come around the ideas of strictness in programming languages again with Typescript and co rapidly gaining in popularity. It&#x27;s just such a massive shame the entire Web community forced this massive wheel-reinventing cycle on us for no damn reason, completely ignoring the lessons learnt -literally everywhere- else.

Custom elements, CSS Namespaces, client side templating, all things that we had decades ago and were just dumped. Abandoning XML and XHTML is and continues to be the webs biggest mistake. People can pretend there were good reasons to stick with HTML but having &quot;On Error Resume Next&quot; baked into your language was a known bad practice even then...<p>I am optimistic for XMLs future though, as the rise of stuff like TypeScript and Rust shows that people do appreciate strict languages nowadays, and WASM means that modern versions of XSLT can be utilized on browsers even if the browser vendors cannot be bothered implementing them.

I loved xHTML, but it failed for a very simple reason: people designing it were working on a DSL to structure a document. But the vast majority of users were actually trying to make things look pretty.<p>Most websites are currently blobs of div. Why ? Because the idea of a web page being mainly a document completely missed the part were artists fiddling with messy code would produce things that customers would like better.<p>It&#x27;s the same reason we currently see giant images on home pages and 10mo static assets to load. Because in the end, the best technical decision doesn&#x27;t win. The one that gives the result the customers end up preferring wins.<p>If you have 3 hours to get a design done, you have the choice between making it look great, or having a clean semantic and maintainable markup, you will have to drop one. And the market will select according to the result.

&gt; &quot;This is a far cry from the systems that were built and proven during the 1970s and 1980s.&quot;<p>I wish the author had enumerated some examples here.

&gt; The current CSS effort provides a style mechanism well suited to the relatively low-level demands of HTML but incapable of supporting the greatly expanded range of rendering techniques made possible by extensible structured markup. The counterpart to XML is a stylesheet programming language that is...Published in April, 1996, DSSSL is the stylesheet language of the future for XML documents.<p>I started building a custom JSON representation of HTML templates with embedded functions for importing fragments and generating markup etc and soon realized I was re-inventing XSLT [0] but kept going anyway. I haven&#x27;t really seen another template language that can translate one DOM tree to another. It seems like XSLT is the eventual implementation of DSSSL in TFA, I&#x27;d be curious if anyone can speak to why it never caught on.<p>[0] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;XSLT" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;XSLT</a>

Going back to the late 90s, I think these are the only interesting bets taken by the industry:<p>- Ericsson&#x27;s Erlang&#x2F;OTP<p>- Sun&#x27;s &quot;The Network is the Computer&quot; and &quot;Utility Computing&quot;, IoT - &quot;Jini&quot; (Internet-connected toasters;), and non-PC devices - smartphones, set-top boxes, smart cards - JavaCard is still popular.<p>- Bell Labs&#x2F;Lucent&#x27;s Plan 9 and Inferno OS (neither is gained adoption)<p>- Semantic Web (which never took off).<p>IMO, XML and Enterprise Java were regression, but so is REST, JSON and dynamic scripting languages. The most recent example is GoLang.<p>EDIT: the XML itself is a good family of specs (especially XSD, XPath and XSLT), the problem was with abusing XML by forcing it onto everything: from configuration to logs to over-the-wire format like in XMPP. I remember IBM even sold XML Accelerator appliances ;)

XML is language-oriented, it&#x27;s like a typical programming notation but without a parsing step (or, rather, with a standard parsing step). It&#x27;s a very unusual thing in this aspect and, I guess, this is the source of all the misunderstanding.<p>E.g. it&#x27;s very successful in things like layout&#x2F;UI description, which is naturally language-like. And not that successful in data exchange applications, which is not language-like. (To exchange data with a language-like medium is like integrating programs by constructing their command lines; possible, but way less convenient than by using a library.)<p>If we are to compare it with something it would be Loglan or Lojban, not JSON. The difference is that it&#x27;s a working and widely used Loglan. Something like this will have to be central to navigate a vast sea of documents, which is what Web was supposed to be. Web has changed quite a bit, but those documents didn&#x27;t go anywhere and the problem is as acute as ever. More acute, I would say; in 1997 there wasn&#x27;t that much content there.

It&#x27;s so easy to look back at XML with ridicule, sitting on the hill of json pragmatism success. But take a moment to consider what we had before XML. Yes, XML was awesome. Back then you&#x27;d consider yourself lucky working with data available in a format that had implementations across multiple operating systems.

XML suffered from the verbosity of having both open and closing tags and the weird uncertainty around attributes vs. child tags, but the real problems came from consultant-driven programming wanting to put XML <i>everywhere</i> and use if for <i>everything</i>. Ant &amp; Maven. SAML. XML for configuration and DI frameworks. *ML everything. Did you know there was a thing called VoxML[1]? Basically XML for phone menus (remember those? Press 1 for English, Press 2 for ... Press 9 to hear these options again).<p>SAX was the hammer applied to every problem.<p>1. <a href="http:&#x2F;&#x2F;xml.coverpages.org&#x2F;voxML.html" rel="nofollow">http:&#x2F;&#x2F;xml.coverpages.org&#x2F;voxML.html</a>

XHTML was my preference, even had some of the semantics I wish weren’t lost (&lt;h&#x2F;&gt;). It’s not a hill I want to die on, but I do very much wish there was a successful push for its strictness.

Seeing this here, I can&#x27;t help but feel that there were some good stuff in the vision for the future of xml. I&#x27;m not saying we should bring back xml, but maybe a few of the ideas should be salvaged.

Idea of XML being superior to anything was the plague nearly everyone at the time caught. Evangelists would be the chosen one or so they thought. But they turned out to be a bunch of con men chasing easy fame. Having once been vaccinated with dynamic programming languages and coding-by-convention, who will ever go back there?<p>Oh... typescript... you...!!!

And then we tried to use it.

Everything XML wanted to be, JSON did better. Xml validation and namespaces were kind of half way there solutions that just made things more complex than they needed to be.

XML was a dumpster fire from day one. It burned hot for a while but thankfully has all but faded away.

XML is the finest example of cargo cult engineering.<p>&quot;The Web was a huge success, there must be a reason. Oh I know, it must have been HTML. We need more of that, and bigger, extensible! Behold, XML!&quot;<p>I remember shaking my head in wonder during the height of the XML hype. What were all those supposedly very smart people thinking?<p>To me, HTML was already a huge mistake, the Web succeeded in spite, not because of it. What could you do in HTML&#x2F;XML that you couldn&#x27;t in s-expression, only simpler, more readable, and more writable? How many man-years have we collectively wasted having to read&#x2F;write&#x2F;generate&#x2F;process&#x2F;validate&#x2F;... the mess that is HTML&#x2F;XML, as opposed to a simpler, more sensible format such as s-expression?