The Future of XML

An anonymous reader writes "How will you use XML in years to come? The wheels of progress turn slowly, but turn they do. The outline of XML's future is becoming clear. The exact timeline is a tad uncertain, but where XML is going isn't. XML's future lies with the Web, and more specifically with Web publishing. 'Word processors, spreadsheets, games, diagramming tools, and more are all migrating into the browser. This trend will only accelerate in the coming year as local storage in Web browsers makes it increasingly possible to work offline. But XML is still firmly grounded in Web 1.0 publishing, and that's still very important.'"

Comments

Much Ado About Nothing...

[milsoRgen]

FTA:

Netscape's dream of replacing the operating system with a browser is also coming true this year.
They've been saying that for years, and frankly it won't happen. A vast amount of users relish the control that having software stored and run locally provides. Of course there will always be exceptions as web based e-mail has shown us.

As far as the future of XML... I can't seem to find anything in this article that states anything more than the obvious, it's on the same path it's been on for quite some time.

FTA:

Success or failure, XML was intended for publishing: books, manuals, and--most important--Web pages.
Is that news to anyone? My understanding of XML is that it's intended use is to provide information, about the information.

Re:"How will you use XML in years to come?"

[DragonWriter]

Sparingly. JSON is just plain better, and doesn't inflict an enterprisey mindset on anyone that tries to use it.

JSON/YAML is/are better (not considering, of course, the variety and maturity of available tools; but then, perhaps, you don't always need most of what is out there in XML tools, either) for lots of things (mostly, the kinds of things TFA notes XML wasn't designed for and often isn't the best choice for),things that aren't marked-up text. Where you actually want an extensible language for text-centric markup, rather than a structured format for interchange of something that isn't marked-up text, XML seems to be a pretty good choice. Of course, for some reason, that seems to be a minority of the uses of XML.

Too many 'this stuff sucks' moments

[Bryan Ischo]

I have had far too many 'this stuff sucks' moments with XML to ever consider using it in any capacity where it is not forced upon me (which unfortunately, it is, with great frequency).

I first heard about XML years ago when it was new, and just the concept sucked to me. A markup language based on the ugly and unwieldy syntax of SGML (from which HTML derives)? We don't need more SGML-alikes, we need fewer, was my thought. This stuff sucks.

Then a while later I actually had to use XML. I read up on its design and features and thought, OK well at least the cool bit is that it has DTDs to describe the specifics of a domain of XML. But then I found out that DTDs are incomplete to the extreme, unable to properly specify large parts of what one should be able to specify with it. And on top of that, DTDs don't even use XML syntax - what the hell? This stuff sucks.

I then found that there were several competing specifications for XML-based replacements for the DTD syntax, and none were well-accepted enough to be considered the standard. So I realized that there was going to be no way to avoid fragmentation and incompatibility in XML schemas. This stuff sucks.

I spent some time reading through supposedly 'human readable' XML documents, and writing some. Both reading and writing XML is incredibly nonsuccinct, error-prone, and time consuming. This stuff sucks.

Finally I had to write some code to read in XML documents and operate on them. I searched around for freely available software libraries that would take care of parsing the XML documents for me. I had to read up on the 'SAX' and 'DOM' models of XML parsing. Both are ridiculously primitive and difficult to work with. This stuff sucks.

Of course I found the most widely distributed, and probably widely used, free XML parser (using the SAX style), expat. It is not re-entrant, because XML syntax is so ridiculously and overly complex that people don't even bother to write re-entrant parsers for it. So you have to dedicate a thread to keeping the stack state for the parser, or read the whole document in one big buffer and pass it to the parser. XML is so unwieldy and stupid that even the best freely available implementations of parsers are lame. This stuff sucks.

Then I got bitten by numerous bugs that occurred because XML has such weak syntax; you can't easily limit the size of elements in a document, for example, either in the DTD (or XML schema replacement) or expat. You just gotta accept that the parser could blow up your program if someone feeds it bad data, because the parser writers couldn't be bothered to put any kind of controls in on this, probably because they were 'thinking XML style', which basically means, not thinking much at all. This stuff sucks.

Finally, my application had poor performance because XML is so slow and bloated to read in as a wire protocol. This stuff sucks.

XML sucks in so many different ways, it's amazing. In fact I cannot think of a single thing that XML does well, or a single aspect of it that couldn't have been better planned from the beginning. I blame the creators of XML, who obviously didn't really have much of a clue.

In summary - XML sucks, and I refuse to use it, and actively fight against it every opportunity I get.

Re:I don't understand...

[machineghost]

The "bulkiness" of XML is also it's strength: XML can be used to markup almost any data imaginable. Now it's true that for most simple two-party exchanges, a simpler format (like comma separated values or YAML or something) would require less characters, and would thus save disk space, transmit faster, etc.

However, the modern programming age is all about sacrificing performance for convenience (this is why virtually no one is using C or C++ to make web apps, and almost everyone is using a significantly poorer performing language like Python or Ruby). We've got powerful computers with tons of RAM and hard drive space, and high-speed internet connections that can transmit vast amounts of data in mere seconds; why waste (valuable programmer) time and energy over-optimizing everything?

Instead, developers choose the option that will make their lives easier. XML is widely known, easily understood, and is human readable. I can send an XML document, without any schema or documentation, to another developer and they'll be able to "grok it". There's also a ton of tools out there for working with XML; if someone sends me a random XML document, I can see it syntax colored in Eclipse or my browser. If someone sends me an XML schema, I can use JAXB to generate Java classes to interact with it. If I need to reformat/convert ANY XML document, I can just whip up an XSLT for it and I'm done.

So yes, other formats offer some benefits. But XML's universality (which does require a bit of bulkiness) makes it a great choice for most types of data one would like to markup and/or transmit.

P.S. JSON is just as usable? Try writing a schema to validate it ... ok I admit, that wasn't so hard, just some Javascript right? But now you have to write a new batch of code to validate the next type of JSON you use. And another for the next, and so on. With XML, you have a choice of not one but four different schema formats; once you learn to use one of them, you can describe a validation schema far more quickly than you ever could in Javascript.

Same deal with transformations: if you want to alter your JSON data in a consistent way, you have to again write custom code every time. Sure XSLT has a learning curve, but once you master it you can accomplish in a few lines of code what any other language would need tens or even hundreds of lines to do.

Re:I don't understand...

[Xtifr]

Like a lot of things, XML is popular because it's popular. Parsing is done with libraries, so programmers don't have to see or care how much overhead is involved, and it's well-known and well-understood, so it's easy to find people who are familiar with it. Every programmer and his dog knows the basics. It's easy to cobble up some in a text editor for testing purposes. You can hand it off to some guy in a completely separate division without worrying that he's going to find it particularly confusing. And you can work with it in pretty much any modern programming language without having to worry about the messy details. It's the path of least resistance. It may not be good, but it's frequently good enough, and that's usually the bottom line.

I mean, yeah, when I was a kid, we all worked in hand-optimized C and assembler, and tried to pack useful information into each bit of storage, but systems were a lot smaller and a lot more expensive back then. These days, I write perl or python scripts that spit out forty bytes of XML to encode a single boolean flag, and it doesn't even faze me. Welcome to the 21st century. :)

Why not S-expressions?

[corsec67]

S-expressions (think the lisp format) are much nicer, more compact, and easier to use than XML, while sharing almost all of the same properties otherwise.

For example:
<tag1>
    <tag2>
      <tag3/>
    </tag2>
<tag1>

becomes:
(tag1
    (tag2
        (tag3)
    )
)

XML is a fad, STEP is the future

[chip2004]

XML tries to make everything fit into a single hierarchy. Most real-world information is comprised of graphs of data. ISO STEP provides better readability compared to XML, a more strongly typed schema mechanism, and a more compact size. Best of all, programs can process and present results of STEP incrementally instead of requiring closing tags so you can hold gigabytes of information in the same file and seek randomly.

Example:
#10=ORGANIZATION('O0001','LKSoft','company');
#11=PRODUCT_DEFINITION_CONTEXT('part definition',#12,'manufacturing');
#12=APPLICATION_CONTEXT('mechanical design');
#13=APPLICATION_PROTOCOL_DEFINITION('','automotive_design',2003,#12);
#14=PRODUCT_DEFINITION('0',$,#15,#11);
#15=PRODUCT_DEFINITION_FORMATION('1',$,#16);
#16=PRODUCT('A0001','Test Part 1','',(#18));
#17=PRODUCT_RELATED_PRODUCT_CATEGORY('part',$,(#16));
#18=PRODUCT_CONTEXT('',#12,'');
#19=APPLIED_ORGANIZATION_ASSIGNMENT(#10,#20,(#16));
#20=ORGANIZATION_ROLE('id owner');

Re:"How will you use XML in years to come?"

[jhol13]

You forgot XSLT.

It is extremely powerful tool, I once (ages ago) made a pure XSLT implementation to convert XML into C. Whith a CSS the XSLT was even browser/human viewable (the output was somewhat similar to the C program output).

I do not think JSON can do that.

Re:XML needs to be easier to read

[Dutch Gun]

I'll save the discussion for XML on the web for others - I'm a game programmer, so I deal with XML as file-based data sources.

At the game studio where I work, all our newest tools are written in C#, and use XML as a data source (typically indirectly though serialized objects). Heavyweight objects (textures, models, audio) are naturally stored in a binary format, which is optimized for the task at hand. The XML-based formats are essentially our game data's source files, and tends to function in a metadata-type capacity. As a simple example, our audio scripts store a lot of parameters about how to play a sound (pitch and volume variations, choosing among multiple variants, category and volume data, etc), and this metadata simply references external binary audio files, typically stored in a standard format like Ogg Vorbis or ADPCM compressed wave files. This metadata is compiled into a binary run-time version using a proprietary format designed to allow us to easily filter versions. These binary formats are then packed into larger containers for simpler management. Since I work on an MMO, we have to think about versioning our binary data, which tends to be challenging.

XML is a great format for us, being so widely supported, since we use both native parsing libraries as well as a lightweight custom parser for our C++ tools (or if we need to support in-game loading for the in-house version of the game). It's easy to look into a file format to see what might be going wrong using just a text editor, and with .NET's reflection capabilities, it's absolutely brainless to easily serialize any data structure. We've decided to use XAML (an XML-based object declaration format) / WPF as well. The artists love the flexibility in the tools, and can even participate in helping to design the interfaces by creating styles.

I don't know what the argument about not knowing what every tag means, like in HTML. The entire point of XML is to be extensible, meaning that it's the client application that determines what the tags ultimately mean. And using SweetXML, btw, misses one of the great benefits of using XML, which is that's it's a standard for which you're likely never going to have to write parsing libraries. It's fine if you want to go that route, but just be aware that you may not have the choice of libraries that you would have by using standard XML.

XML does tend to suffer from the "golden hammer" syndrome. Honestly, I'm not a huge fan of it's verbosity or general readability either, but if you take it for what it is, and use it sensibly, it's just another nifty tool you as a programmer can make good use of. After all, wouldn't you rather be working on more important parts of your project than fiddling with a text parser?

Make working with XML suck less...

[scottsevertson]

"XML is really just data dressed up as a hooker."

--Dave Thomas

XML does suck if you stick with some of the W3C standards and common tools. Suggestions to make it less painful:

• Ditch W3C's XML Schema

  W3C Schema is painful; it forces object-oriented design concepts onto a hierarchical data model. Consider RELAX NG (an Oasis-approved standard) instead; it's delightful in comparison. Use the verbose XML syntax when communicating with the less technical - if you've seen XML before, it's pretty easy to comprehend:

  <r:optional>
  <r:element name="w3cSchemaDescription">
  <r:choice>
  <r:value>painful</r:value>
  <r:value>ugly</r:value>
  <r:value>inflexible</r:value>
  </r:choice>
  </r:element>
  </r:optional>

  Switch to the compact syntax when you're among geeks:

  element w3cSchemaDescription { "painful" | "ugly" | "inflexible" }?

  There's validation support on major platforms, and even a tool (Trang) to convert between verbose/compact formats, and output to DTD and W3C Schemas. And, if you need to specify data types, it borrows the one technology W3C Schema got right: the Datatypes library.

• Don't use the W3C DOM

  The W3C DOM attempts to be a universal API, which means it must conform to the lowest common denominator in the programming languages it targets. Consider the NodeList interface:

  interface NodeList {
  Node item(in unsigned long index);
  readonly attribute unsigned long length;
  };

  While similar to the native list/collection/array interfaces most languages provide, it's not an exact match. So, DOM implementers create an object that doesn't work quite like any other collection on the platform. In Java, this means writing:

  for(int i = 0; i < nodeList.length(); i++)
  {
  Node node = nodeList.item(i);
  // Do something with node here...
  }

  Instead of:

  for(Node node : nodeList)
  {
  // Do something with node here...
  }

  Dynamic languages allow an even more concise syntax. Consider this Ruby builder code to build a trivial XML document:

  x.date {
  x.year "2006"
  x.month "01"
  x.day "01"
  }

  I thought about writing the W3C DOM equivalent of the above, but I'm not feeling masochistic tonight. Sorry.

  The alternatives depend on your programming language, but plenty of choices exist for DOM-style traversal/manipulation.

• Forget document models entirely (maybe)

  In-memory object models of large XML document can consume a lot of resources, but often, you only need part of the data. Consider using an XMLPull or StAX parser instead. Pull means you control the document traversal, only descending into (and fully parsing) sections of the XML that are of interest. SAX based parsers have equivalent capabilities, but the programming model is uncomfortable for many developers.

  Even better, some Pull processors are wicked fast, even when using them to construct a DOM. In Winter 2006, I benchmarked an XML-heavy application, and found WoodStox to be an order of magnitude faster at constructing thousands of small DOM4J documents