This page is to list what developers want or comments
The ICE Core Taverna (from the SAMPO people) is very nice. It has Python processors. It also is able to do "clever" input browsing by knowing the semantic type of the input and only showing those. You can also save "input sets" either together in a directory or specified by a file.
They do GOTOs - yes GOTOs. The Freefluo engine was altered to support this. It works because the parameters to a service's functionality are changed by other services. The GOTO can be automatic or, I think, forced by the user.
Spent some time with Kevin Benson, of AstroGrid, going through in depth what he is required to do to create a new Activity plugin within the workbench. I explained that there are going to be some changes with 2.1 but he was content with this. His main area of concern was support for optional ports with default values - something that T2 doesn't support. Automatic upstream String Constants didn't meet his needs. We need to concider how we can support default values, and follow up with Kevin in the very near future to see how he's got on with his plugin.
Spoke with Eddie about the code restructuring, which he was pleased about, but is keen to be able to some tidying up of the Biomoby code - so needs notifying as soon as the restructuring is in place. There are not specific bugs, but whilst reviewing his code he felt unhappy about the general structure and needs a bit of maintenance.
Also Eddie is keen for a 1.8 release of Taverna 1 workbench, as there have been Biomoby processor improvements. An 126.96.36.199 update of the Biomoby Processor would be a quick way of providing this.
There was interest in provenance and a provenance service API that goes beyond SPARQL queries over an OPM. A provenance API as an additional t2platform service should be considered. We have use-cases from a User perspective but we also need to think about Developer access to the provenance (which we have been doing, but this need was emphasized at the workshop).
Morris Swertz was interested in automating adding XMLSplitters to the most primative types. Fortunately this should be straight-forward as there is an Edit for it. However, we gave this a go using t2platform and the Edit didn't work as expected, and according to the unit test code. There has been an API change as part of t2platform and I need to check if anything has been broken by this change.