Microsoft asks us to define terms for them; we happily oblige
October 16th, 2008 by Alex Neihaus
In an article by Paul Krill of InfoWorld on Microsoft’s “Oslo” — a sad pastiche of a BPMS — Robert Wahbe, Microsoft’s corporate vice president of the Connected Systems Division, responds to our Michael Rowley’s assertion that a meta-meta-language is for “‘über-geeks” by saying:
“I don’t know what ‘über-geek’ is, but there’s a large class of developers who build apps for a living. Oslo is interesting to such developers because it aspires to raise the level of abstraction and make them more productive and applications more flexible.”
Well, you won’t be surprised to know that we’re happy to define an über-geek for Mr. Wahbe. It’s the guy sitting in the cube marveling at all the abstractions of a modeling-language-for-making-modeling-languages but who is producing absolutely nothing for the business. He or she can’t figure out what to do with the language-to-make-a-language, no matter how hard he or she tries. He or she might even wear something like the watch pictured above…in the vain hope that one day it’ll tell time at a glance.
I can’t figure out if Microsoft is making things complicated for SOA developers simply because they can due to dominance in their developer community or because they really think internal Mircosoft engineering science projects make for interesting, useful, deployable real-world products.
No matter, the real objective isn’t just to cow their developers. It’s also to try…wait for it…to put a dagger into the heart of BPMN and BPEL. Woe unto Microsoft should any of their developers learn a transportable skill. Woe unto their future should enterprises really have any choice for creating BPM applications.
Why else would BPMN and BPEL be simply ”accommodated” in Oslo? (That’s Wahbe’s word for it, not mine.)
Isn’t it just a short step from “accommodation” to “embrace and extend?” Let’s face it, despite joining the relevant standards committees of late, Microsoft has no track record of doing anything that would indicate real, true dedication to standards. I have no doubt the people from Microsoft on these committees are sincere. But they have an internal challenge back in Redmond equivalent to making English the official language of Quebec.
In the US, there’s this old political saying that the Social Security system is the “third rail” of national politics: politicians touch it and die. In the Microsoft world, developers are the third rail of their corporate developer ecosystem: anything — including and especially standards — that touches it must die.
Or be “accommodated.”
Tags: definitions, microsoft, uber-geek
November 6th, 2008 at 4:08 pm
I think you are right, BPEL and BPMN will die as intermediate formats. “Oslo” will take us from sharing angel brackets to sharing models in a common repository through a common meta language “M”.
November 6th, 2008 at 4:48 pm
Hmmm….I think we said just the opposite: that a common meta language is a seriously bad idea, one that Microsoft is foisting on the world largely because in its own internal logic, M is “neat” or “cool” and not because anyone wants a language to define languages.
Still, Casper, if you think this is what you want, may the Force be with you as you try to actually make something work on a real computer in a real application.
November 8th, 2008 at 10:53 am
Thanks for correcting my standpoint, Alex
Alex, it’s not me wanting a shared repository – it’s what happening in this very moment. I can’t say I’m blown away about the market acceptance for BPEL in last 5 years. “Oslo” will play the role for building applications as Windows has been playing for the last 2 decades. You know it, I know it and all vendors know it’s time to let the standardization of how things execute go away. When it comes to portability reasons, BPEL is just a joke. Don’t get me wrong though. I like BPEL, but just it’s semantics for process execution – as much as I hate the fuzzy, unlogical notation of BPMN.
November 10th, 2008 at 9:22 am
Casper — actually BPEL portability is not a joke. It _is_ possible (and not even very hard) to stick to the standard when creating orchestrations. If you do, they will work in ActiveVOS or in Oracle’s engine or IBM’s engine. There used to be a problem that human activities weren’t supported, but that is being solved by BPEL4People.