Release notes for the 'Christmas' release of the Taverna Workbench, v1.7.0. This release contains the first technology preview of the new Taverna 2 (T2) codebase in the form of a plugin to translate a version 1.x workflow and run in the new enactor.

See myGrid JIRA issue tracking for more detail at http://www.mygrid.org.uk/jira. The following outstanding issues with older releases have been fixed in this one :

This set of changes and fixes should make it easier for plugin authors depending on system level artifacts (such as the next Soaplab2 processor). There are relatively few new features in the version one code because we've been focusing on...

The 1.7.0 release of the Taverna Workbench incorporates a technology preview of the new Taverna 2 (henceforth T2) enactment engine. This is not the full T2 release scheduled for June 2008, neither should it be regarded as suitable for production usage. In this release we have implemented the T2 functionality as a plug-in to the standard Taverna Workbench. This means that if you are not interested in the new enactor you can safely ignore it – the workbench will run exactly as normal and you can simply not use the extra perspective the new enactor plug-in will have created; the default ‘run workflow’ option will use the version 1.7 enactor.

The version of the T2 enactor in this release has certain limitations; as the underlying model manipulated by the workbench is still the Taverna 1.x functionality such as staged iteration and configuration of the dispatch stack (used to define recursive behaviours, dynamic binding to abstract resources and similar) cannot be accessed (although they are implemented and usage of T2 from its API provides access to them).

As the plug-in works on a ‘translate from T1 and run’ model we can only make use of features in T2 which are inherited from the T1 model – you could run a workflow with staged iteration, for example, but there’s no way to specify it in the UI at the moment. There are also limitations in terms of the available processor types. For each processor class (web service, biomart etc) in Taverna 1 we have to implement a translator and an equivalent activity class from the T2 API. Not all processor types are translated, specifically the API consumer and BioMoby types as well as any third party extensions of which we are unaware.

The majority of the local worker types are translated, so workflows which make use of web services, local workers, beanshell scripts, biomart and soaplab should be translated without any issues. These limitations notwithstanding, the new enactor is capable of scaling far beyond the old version 1 series both in terms of memory usage and performance. The former is enabled by a pass-by-reference model, the latter through pipelining.

In addition we have an enhanced monitoring facility which allows the state of nested workflow instances to be exposed and in this release we have implemented a naïve user interface to this functionality in the form of a monitor tree. There is also enhanced error checking, something that will extend throughout the new codebase and which allows us to catch potential problems in the workflow model before they occur, ideally before the workflow is run.

In summary – the T2 functionality in this release of the workbench is intended to give you some insight into the potential capabilities of the full T2 release scheduled for June of next year. Bearing in mind the caveats above please do try to run your workflows in the new enactor and let us know how you get on – we are particularly concerned with workflows which fail where they worked or which return different results. There may be legitimate reasons for both these cases but we would like to know about them. We anticipate one more release where this T1 / T2 split exists sometime around the end of March next year before migrating across to the full T2 version in June. We should point out that the full T2 workbench will be capable of loading your current workflows, one of the purposes of this six month shakedown period is to ensure that this transition can take place with minimum disruption to your existing work – we understand and value the time you have spent creating it and have no intention of disregarding it.

We're currently writing up documentation on the core architecture of Taverna 2 (as this is now sufficiently stable to be useful!), this will be made available online and notifications published to the mailing lists early next year.

This release also includes a prototype of the new activity palette (activity is the T2 version of processor, to a reasonable approximation), described in its own section in the user manual for this release. As with the new enactor the aim is to get feedback rather than be ready for mainstream production use.

A guide to using the Activity Palette can be found at http://www.mygrid.org.uk/usermanual1.7/t2_guide.html#activity_palette.