XSLT is in decline - what's replacing it?

Question

According to Google trends, XSLT is declining fast: http://www.google.com/trends?q=xslt If we extrapolate it as straight line, it will reach zero in 3 years (though it's probably curving, so it will asymptote to zero).

XML itself is also declining, at a similar rate, but seems to have leveled off a lot already: http://www.google.com/trends?q=xml

1. Is it simply that XML is used less, and so XSLT is also used less? e.g. the tasks formerly performed by XML are now done in other ways (like JSON for webapps).

2. Perhaps it's just that XSLT has stopped growing, and most XSLT users have become experts by now, who don't need to google as much to learn about it? Or perhaps they do it within another tool (like XMLSpy), which has adequate documentation built in?

3. Or is it that the transformation tasks that XSLT performed have been absorbed into other products, and so become invisible to most users?

What tools are now being used to transform XML data, instead of XSLT?

Answer 1

In the case of XML, I would say that it's become less of a buzzword, and more accepted as a normal part of software development, so it's googled for less often.

In the case of XSLT, I think that's declining fast because it didn't live up to expectations (it certainly didn't live up to mine!). I don't think transforming one kind of XML into another has turned out to be quite the silver bullet that people thought it might be.

Answer 2

I doubt anything is replacing it, but rather that it never quite caught on. It is a superb technology, but unfortunately the web is crap and is largely incompatible with most of the benefits offered by XML derived technologies. As a result I can only guess such awesome concepts are slowing being abandoned in favor of more popular and inefficient means.

Relax NG is a schema technology, which does not serve to describe transformations very effectively.

Answer 3

I found that XSL-T was cryptic and difficult. I used it infrequently, and every time I did I had to "relearn" it again.

It's easy to write spectacularly bloated, slow XSL-T scripts. Whenever I'd see lots of "dot" notation to navigate up and down the tree, instead of using a more natural matching and top down approach, I knew the author was in trouble.

Mostly, I think that templating solutions like Velocity and XPath are far more intuitive and easy to understand. XSL-T won't entirely disappear, because it does have its uses, but hopefully we'll see an end to its overuse.

Answer 4

In .NET a lot of transformations can be done easily with LINQ to XML. I've found in the past where I might of used XSLT I now use LINQ to XML.

I find C# programmers can pick up LINQ to XML a lot quicker than they can pick up XSLT (no surprise there). That said XSLT still has it's place. Microsoft has a comparison here.

Answer 5

I don't think it is possible to reason out of Google trends. According to this metrics, Java is also declining. Perhaps it's just that the total number of search is not declining at all, but that the fraction of technical searches went down as more and more people search for other stuff related to their lifes in the internet?

Answer 6

I quite like XSLT, but the fact is that a lot of people find it awkward. Additionally, 1.0 has a lot of annoyances (no case-insensitive / regex), and 2.0 is poorly supported and tool too long to arrive.

At one point it was quite popular as a tool to transform xml (as a model) to (for example) a web page - but tools have improved - see the V in ASP.NET MVC as just one example (the tooling for other platforms is better too) - or LINQ-to-XML.

XSLT 1.0 will remain a handy tool in my bag, but I don't think I'm going to invest learning 2.0 unless I absolutely have to.

Answer 7

Well, XSLT is about transforming XML files. However, there are more and more language-specific solutions that offer similar functionality. Most importantly, with .NET development, many programmers can now use Linq, which is similar to XSLT in functionality, namely reorganizing, filtering and ordering a structured list of data. I've noticed this in my own work even! For one project I had to deal with a list of products by company. So first I exported the tables to an XML file, then used a stylesheet to restructure the data. The result was then shown on-screen. But now I don't need the export/import to xml step anymore. All you need is to define a good Linq query to get the same result.

This is interesting since many Visual Studio projects will use a lot of XML data. And Microsoft is doing a lot to hide these stylesheet transformations or to make them just obsolete.

Another reason is -of course- that more and more people are slowly becoming experts at this technique. In 2004, the use of XSLT was still reasonable new and thus many people still had a lot to learn. Now, they've become experts. Take a look at http://www.google.com/trends?q=c# and you'll see that even C# is slowly declining... Same with PHP and Python. It's just that more and more people start to become familiar with these topics, thus they have to search a lot less.

Interestingly enough, the trent for porn is still rising... So is this: http://www.google.com/trends?q=flu And why has the flu suddenly become more interesting? Because of the Swine flu that's supposedly going to make a few billion people sick. Once the main panic is gone and more people learn about it, the interest in the topic is becoming a lot less.