This document describes the format for representation of statistical data that is destined for the Hyades statistical model. Any data that complies to this format will be able to be consumed by Hyades and made available within the tooling. ********************************************************************* Change History: 22 Oct 2003 - Richard K Duggan - initial implementation 10 Dec 2003 - Marius Slavescu - fixed some structural problems ********************************************************************* Copyright 2003 Eclipse.org http://www.eclipse.org/legal/cpl-v10.html This is the base type for all descriptors. DescriptorType's do not have associated counters but rather provide containment structure so that descriptors have a hierarchy. The name of this descriptor as it will be represented in the UI. This leads to some interesting internationalization issues that need to be discussed at a greater level. This is the identification of this particular descriptor. Identifiers are used so that the heirarchy of descriptors can be be provided without having a structured event requiring all the heirarchy to be declared at the same point in time. The identifier of this decriptors parent. If a parent is not provided then this descriptor is considered to be a root element. A DescriptorMember is a descriptor that has a counter associated with it. DescriptorMembers are refered to be Observations. DescriptorMember's contain representation information for the observation information they describe. This is a concrete type of MemberDesciptors that is to be used for descriptors which will have associated observations that are counters. Counters are interpretted as the current value is considered valid at the time the observation was taken. This is a concrete type of MemberDesciptors that is to be used for descriptors which will have associated observations that are samples. Samples are interpretted as the current value is considered updated only on certain intervals. This is the base type of all observations. Observations have an associated MemberDescriptor and a time at which the observation was taken. A concrete observation that contain a textual value. A concrete observation that contains a integer value. A observation that contains a floating point value. The base of all representations. A representation that has a minimun and a maximum value. That is, the observations will reside between these two values. Extends the range representation by adding a minumun and maximum threashold for the observations. The idea here is that you could store these thresholds in the model and then have logic that informs you when an observation is outside of these threshold values. This is the root tag that allows the packaging of many of the predefined fragments into a single file.