I just read on the Web about the availability of this package from Microsoft. Cool. I knew a CTP was coming but didn’t know when.
I’ve been trying to explain to scientific workflow people that Windows Workflow is a generic framework without any built-in semantics and that any type of dataflow/workflow could be implemented, if one invested in building the required semantics as activities. That’s how I had enabled WCF–WF integration for an internal project. The nice CSD (Connected System Division) folks now have a package to support that integration (it’s part of the latest Orcas CTP I believe). My prototype went a step further to enable the automatic extraction of a service’s (implemented as a WF workflow) multi-message interaction contract (described in XAML). Based on that contract, the generation of a consuming WF workflow could be automated (capturing all the possible conversation states with the target service).
Anyway… back to WF. I have been toying with the idea of a Scufl (think scientific workflow) package on top of WF. This way, all those Taverna workflows could be executed on the WF platform… you know, nice debugging, nice visual designer and tooling support, etc. Carole promised to send me the semantic descriptions (captured in Scheme I believe 🙂 of the Scufl language. That’d be an interesting side project. I am seeing Carole next week so I’ll talk to her again about this.
2 responses to “BPEL for Windows Workflow Foundation March CTP”
Hi Sava,
I am not familiar with the WF framework you talk about but I have recently delved into the Taverna workflows and their Scufl descriptions. Transforming the Scufl workflows into BPEL would be easy in theory but the problem I see is that in a Taverna workflow the different activities (processors in the Scufl terminology) can be very different: in addition to the “standard” Web Services, you can also have Biomody services, Biomart services, local processors (e.g. for concatenating strings), etc. Workflows comprised by these non-WS activities will require additional machinery and wrappers to be transformed into BPEL.
Taverna seems to be a de facto standard in bioinformatics but trying to be everything to everybody and integrating all existent tools and services put some difficulties to its further evolution.
Cheers,
Stelios
Hi Stelio,
Thank you for your comment.
Please note that i was not suggesting that Scufl be converted to BPEL. I gave the BPEL package for WF as an example of the implementation of any framework’s semantics ontop of the WF platform. One could think that the Scufl semantics could be implemented as WF “activities” (the concept used in WF to capture the processing nodes of a graph).
Each graph node in Scufl has some specific semantics associated with it. Those processing semantics could be implemented as activities and so a translation from Scufl to WF XAML (the markup used) would be trivial.
BTW… I am aware of an effort at Indiana University to convert Scufl (I think) to BPEL. Beth Plale and Dennis Gannon are going to be visiting us this coming week so I’ll ask them if we get the opportunity. In fact, I am having dinner with them on Wednesday 🙂