2009/3/31 Nick Gall <[email protected]>: > On Mon, Mar 30, 2009 at 7:10 PM, Steve Jones <[email protected]> wrote: >> Whether it is individuals (Bjarne, Bertrand (Meyer), Roy or Gosling) >> who come up with a new technical approach or companies who seek to >> sustain the status quo the end result is fundamentally the same, there >> are some great advances and the bar is raised but the day to day >> building from that point remains poor. > > However "poor" technology leaps are, they are infinitely better than > hand-waving pseudo-architectural leaps like SOA.
I disagree on two points, firstly that SOA is "hand-waving" and secondly that technological leaps are "infinitely" better. I'd argue that methodology leaps, OO, SOA, Iterative development (which you may consider hand-waving) have had a much bigger impact than the specific technology elements. > One guru you left off your > list is Tim Berners-Lee. He started the Web, and we've all been the better > for it. The man is brilliant, and his brilliance was in the W3C, one man started it and a committee codified it. IMO that is the way to go. Java nearly got there with the JCP but then Sun started subverting (IMO) the process. > Same for Licklider/DARPA/BBN and the Internet. For that matter, same > for Fred Brooks and OS/360. I wouldn't say that DARPA is an individual or that Brooks was the single guiding leader behind OS/360. There are loads of brilliant minds in the short history of IT (only one genius though, Alan Turning) but the question was around a consistent single leader being a good thing. I don't think TBL would argue that he is the single consistent leader of WWW. > > For better or worse, it is concrete technologies that move the human race > "forward"; not vague "architectures". Human race? Getting a little bit high and mighty. I'd again say that at the "human race" level it is "vague" (i.e. non-concrete things) that move us on more, democracy being a good example. > REST is to SOA as the Web is to > Hypertext. There were/are lots of hypertext gurus who dreamed of perfect > architectures (Ted Nelson comes to mind), but never implemented anything > popular enough to move forward. TBL came along and delivered a wildly > popular "dumbed-down" version of hypertext (eg no backlinks - heresy at the > time and still heresy to Ted). And the rest is history. Same goes for SOA: > lots of theory, but no wildly popular technology realization -- except REST. REST != WWW. WWW != REST The PRINCIPLES of SOA in terms of service orientation are (IMO) the powerful elements as they enable organisations to _structurally change_ the way that they work and it is through those structural changes that actual progress is made rather than through technologies (Mythical Man Month stuff). The thought process remains the hardest part at it is therefore conceptual models that have the most ability to improve IT as they can impact ALL of IT rather than simply its implementation. REST works in some places, it doesn't work in all places. The internet is a small part of enterprise IT and legacy continues, and will continue, to dominate. For this reason it is about the conceptual models that enable us to evolve and adapt those IT estates that will have the biggest impact rather than the latest generation of lipstick on the pig technologies no matter how good anyone feels those technologies are today. > > Gall's law wins again: "A complex system that works is invariably found to > have evolved from a simple system that worked. The inverse proposition also > appears to be true: A complex system designed from scratch never works and > cannot be made to work. You have to start over, beginning with a working > simple system." Read "REST" for "simple system" and "SOA" or "WS-*" for > "complex system". SOA != WS-* in the same way as REST != WWW. REST is NOT a system, its a whitepaper and an software design approach for systems, "hand-waving" to use your language. SAP and Oracle integration are two examples of how your "complex" view doesn't match up against what can now be done over the previous integration approaches and WS-* is absolutely an evolution over previous approaches according to those, including IIRC yourself, who claimed it was just an RPC approach. Gall's law is actually just Church-Turing expressed in a lot more words BTW. Steve > -- Nick >
