Jan K. Labanowski: Computational Portals for Chemistry  

Application and Task Descriptors

Before any application can be executed, we need to know:
  • how it can be run (e.g., path to the executable, parameters)
  • where the application runs (e.g., machine, queues)
  • what it needs to run (data files, their format, type, names)
  • what it produces (files -- format and type, logs, stdout, etc.)

The structure which keeps this information is a called Abstract Application Descriptor (AAD) -- an XML file and the corresponding DTD. Once we are about to execute the application, we need to convert some of the generic requirements into actual run-time parameters (e.g., actual file names, actual queue, etc.) and we get Application Descriptor (AD). The details can still be found at the original NPAC site: http://www.npac.syr.edu/users/haupt/WebFlow/MODULES/AAD.html or locally here.

The applications have to be combined together to form a computational session. Relation between individual applications, which are combined to form the session, is described by Abstract Task Descriptor (ATD). This is again an XML file and lists pointers to Application Descriptors or other tasks (i.e., the groups of applications working together). For the description, ATD's DTD, and examples check http://www.npac.syr.edu/users/haupt/WebFlow/MODULES/ATD.html or see the local copy here. Their format and role is also discussed in http://www.npac.syr.edu/users/haupt/WebFlow/XML/model.html or its local copy: here.