 Sunday, March 28, 2004
|
|
|
|
The always astute Phil Wainewright notes a specification called BPELJ, or Business Process Execution Language for Java.
Now the XML specification BPEL (sans the J) is an exciting new way of codifying and automating complex business processes. Many including this writer have claimed that by incorporating for the first time a temporal component it represents an entirely new programming model.
But now IBM and BEA have sought to co-opt the XML standard by enabling procedural Java to be mixed in. Phil rightly calls it a "sleazy compromise" which glorifies "mongrel code." First, the spec attempts to crown Java as the only procedural language worthy of inclusion in a BPEL spec; and, worse, IBM and BEA find they have to change the architecturally neutral BPEL spec in order to support Java. And then they offer that sometimes you mix Java in your BPEL file, but other times you mix BPEL fragments in your Java file. Amazing! It's a clear case of corrupting an open standard with closed technologies.
Worst of all, however, is the very notion of intermingling declarative and procedural programming models. To me this violates every advance we have made in software engineering for the last forty years: from structured to object-oriented to service-oriented programming. If nothing else we've learned how to layer code in appropriate ways, but BPELJ seems to blithely ignore any attempt at functional isolation.
|
|
|
|
March 2004 |
Sun |
Mon |
Tue |
Wed |
Thu |
Fri |
Sat |
| |
|
|
|
|
|
6 |
7 |
|
|
|
|
12 |
13 |
|
15 |
|
|
|
19 |
20 |
21 |
|
|
24 |
|
26 |
27 |
28 |
29 |
|
|
|
Feb Apr
|
