This chapter is from the book
This chapter is from the book
This chapter is from the book
The Birth of Mashups
- You can have it "good," "fast," or "cheap." Pick any two of the three.
- —Classic programmer's adage
Quick, easy, and affordable application development has always been a goal of software engineering. Reusing something that's already been built, tested, and paid for is one of the quickest ways to achieve this objective. From subroutines, to external libraries, to object orientation, to templates, to Web Services, each great advance in programming has been born from the desire to reuse material instead of starting from scratch. The limitation inherent in each of these milestones is that they were created by developers for the sole use by others in their profession.
It seemed inevitable that with the vast amount of new material being placed on the Web 2.0 every second, it could somehow evolve into raw material for software development. Tim Berners-Lee envisioned this leap in Web reusability in what he termed "the semantic Web," which describes a platform for the universal exchange of data, knowledge, and meaning.11 And while work continues to define new languages and protocols to realize Sir Tim's dream, mashups are making this vision a reality now.
Mashups are an empowering technology. In the past, resources had to be designed for reuse. Application program interfaces (APIs) had to be created, packages compiled, documentation written. The application developers and solution architects who recycled resources were subject to the whims of the original designers. With mashups, you aren't limited to reusing an existing API; you can impose your own if none exists. So if an application or site offers no API, or if you don't like the access methods that are already in place, you can design and implement your own (see the API Enabler pattern in Chapter 4 for several examples). The promise of achieving programmatic access to almost unlimited data is intoxicating. Even more exciting is the notion that the tools for constructing mashups have begun to reach a level of usability where even nontechnical users can build their own solutions.
Many popular definitions of a mashup would have you believe the term is limited to a combination of Web-based artifacts: published APIs, RSS/Atom feeds (see the "RSS and Atom" sidebar), and HTML "screen scraping." Although there are certainly valuable solutions in that space, a broader world of data can be mashed up, including databases, binary formats (such as Excel and PDF), XML, delimited text files, and more. The rush of vendors attempting to capitalize on the burgeoning market for enterprise solutions hasn't helped bring clarity to the field. To turn a classic phrase on its head, we have a ton of nails out there, and everyone is trying to tell us that they have the best hammer.
Another common misconception is that mashups combine at least two disparate sites to form a brand-new "composite" application, complete with a neat new user interface. That's certainly possible, but mashups need not be an end in themselves. It is more accurate to say that all composite applications are mashups, but not all mashups are composite applications. The enterprise mashup creator can use the technology to transform the Web into his or her own private information source. This data can be used for strategic planning or analysis in systems like Excel or MATLAB. Mashups may also be used to access a single resource at superhuman levels to mine data or migrate content. Creating mashups is all about finding data, functionality, and services and using them to both solve problems and create opportunities.12