PPS User Guide

Use the command Preferences dialog pages to set how you want Eclipse to operate. The menu path is Window  Preferences on Windows and Linux - on macOS the Preferences can be found in the App menu. The filter input at the left top can be used to easily navigate to the PPS settings.

The tmsct file extension is associated with the TMSC textual editor.

The runtime behavior and timing of software applications can be visualized using Timed Message Sequence Charts (TMSC). In a TMSC, executors are shown on as horizontal swimlanes, and (real) time is shown along the horizontal axis. The labels of the executors are shown on the vertical axis. An executor is a single threaded computational resource with run-to-completion semantics. The function calls that are executed by an executor are shown as bars within its swimlane and are called executions. The bars are labeled with the function name. Nested function call executions are drawn as stacked bars, resulting in a call stack. Arrows between the bars denote the data and control flow dependencies. The arrows indicate where executors are communicating with each other. Section Relations to (Message) Sequence Charts and Flame Graphs compares TMSCs to similar representations.

Section TMSC view shows an example visualization.

The tmsc and tmscz file extensions are associated with the TMSC graphical viewer. These files support the following contextual actions:

The Outline view shows the model in a tree view. Clicking an element in the tree selects the element and shows the details of that element in the Properties view.

By selecting elements in the tree both the resource and time context of an editor view can be reduced, see also Section dynamic inclusion filtering. With reduction of the resource context is meant that the editor area will show only content which is related to the selected Outline elements. For example, when a life-line is selected, the TMSC view view will only show this life-line and the life-lines that it directly communicates with. This feature is active if the Push Context toggle bottom is selected.

When the Push Time Range toggle button is selected, the editor area will automatically show the time range which includes the time context for the selected Outline elements.

The model elements in the content outline tree support the following contextual actions:

The Selection view tab is similar to the Outline view. It shows the model in a tree view. Clicking an element in the tree selects the element and shows the details of that element in the Properties view.

This view supports dynamic filtering based on the Outline view selection (see section dynamic inclusion filtering) to filter elements of interest.

In a TMSC, the life-line of an executor is shown on as a horizontal swimlane, and (real) time is shown along the horizontal axis. The executor labels are shown on the vertical axis. The function calls that are executed by an executor are shown as bars within the swimlane. The bars are labeled with their execution name. Nested function call executions are drawn as stacked bars, resulting in a call stack. Arrows between the bars denote the data and/or control flow dependencies. The arrows indicate where executors are communicating with each other.

Due to dynamic inclusion filtering some executors might not be shown; however, the visible executors might communicate with this hidden executors. Communication text balloons are shown in this case, depending on the selected dependencies visibility. The arrow from/to the balloon indicates the data flow of the communication and the text in the balloon describes the other side.

It is possible to select executions or dependencies by clicking on them, also see section Selection in Chart. Selected elements are shown with a thicker yellow border. Once selected, the details of the execution can be seen in the Properties view and in the Data Analysis view.

The Timed message sequence chart view tab supports some common functions to support analysis of its data, like:

On top of these common functions, the TMSC tab has the following additional action bar buttons.

This view can be used to perform analysis on TMSC elements like executions, messages or metrics. By selecting an element, it can be analyzed in detail. All comparable occurrences of the element will be collected and analyzed in the selected chart. What makes an element comparable, depends on the type of the element. For executions, their function and either component or executor are considered.

Apart from the standard analysis of a single element, there is also the possibility to analyze measurements. To access this function, the user must first perform a time measurement for two executions. By selecting the measurement, the execution pair can be analyzed in detail. All occurrences of execution pairs for the selected measurement will be collected and analyzed in the selected chart.

The charts in the Data Analysis view support some common functions to support analysis of their data, like:

On top of these common functions, the Data Analysis view has the following additional action bar buttons.

The chart viewers supports zooming and panning, as follows:

Scroll up to zoom in and scroll down to zoom out. The location of the mouse will be used as focus area, meaning that the chart will be zoomed in or out around that point.

First click somewhere in the chart to activate keyboard focus. While holding the Ctrl key, the arrow keys can be used to zoom the chart in or out. Both the normal arrow keys as well as the arrow keys on the numpad can be used.

In addition to the arrow keys, also 'browser style' zooming is supported.
While holding the Ctrl key:

Hold the Alt key while clicking and dragging the chart in the direction to pan.

First click somewhere in the chart to activate keyboard focus. The arrow keys can be used to pan the chart in the desired direction. Both the normal arrow keys as well as the arrow keys on the numpad can be used.

Right click on the chart to show a menu with even more options.

Furthermore, the action bar for time based views show additional buttons to zoom.

It is possible to select some of the items in a chart by clicking on them. If an item supports selection and is selected, it is shown with a thicker yellow border. Once selected, the details of the item can be seen in the Properties view.

A context menu is available to select the current chart selection in the Outline view. This might be helpful for views that support dynamic inclusion filtering.

It is possible to measure the time between for instance the start of one task, and the termination of another task in a chart. While hovering over a task, a gray vertical support line is shown. While hovering over the first half of the task, it is shown at the creation of the task. While hovering over the second half of the task, it is shown at the termination of the task. Hold down the Alt key while such a support line is shown, and left click it, to keep the support line fixed at that moment in time. Then hover to move the support line to another end point of a task and again left click while holding the Alt key. An annotation is then shown that visually and textually shows the time between the two support lines. The measurement is also added to the model and as shown in the Outline view and the dirty (*) state of the editor. The user can choose to save the measurement or to undo the operation by using the Edit  Undo menu.

It is possible to save a snapshot of a chart by clicking the Export As Image from the view menu (small triangle or 3-dots button at the right top of the view). The action will open a save dialog which allows you to select the location and the name of the exported PNG file.

The time range in open viewers can be synchronized, so that multiple aspects on a time range as shown in the different viewers, can be analyzed simultaneously. When one editor is zoomed or scrolled, the other editors remain synchronized. This is especially helpful when editors are positioned above or 'on top of' each other. As in that case, the margins to the left and right of the time range will be synchronized as well, such that the time ranges have the same width in the viewers. The data for corresponding points in time are then vertically aligned, making it easy to correlate the data from the different viewers.

Only PPS viewers that have a time range (typically the x-axis) support time range and margin width (the width taken by the margin labels, margin ticks, etc) synchronization. If a viewer supports time range and margin width synchronization, some buttons will be present in the toolbar of the viewer.

Currently, PPS viewers do not support manually changing the margin width. As such, the margin width is only synchronized by explicit request, using the Synchronize Margin Width toolbar button (). To reset the margin width of an editor/view, click the Refresh toolbar button ().

Disabling time range and margin width synchronization can be useful when dealing with large traces. Zooming in to a specific part of a trace while having time range synchronization enabled may lead to performance issues. To mitigate this, disable time range synchronization on a view, zoom in to the time range of interest within that view, re-enable time range synchronization for the view, and press the Synchronize Time Range toolbar button () to explicitly synchronize the new time range to the other views. That way, the other views are only updated once.

Time range and margin width synchronization are performed per workbench window. If you open a second workbench window, the editors/views within each workbench window are synchronized. Editors/views across different workbench windows are not synchronized. This is helpful when looking at different traces in separate workbench windows.

Some editors have multiple tabs, where each tab has a different view on the same data. It can be useful to have multiple editors on the same data, to show different views on the same data, at the same time. To support this, editors can be split and cloned.

When an editor has the focus, the following toolbar buttons are present in the toolbar:

When splitting, the two editors remain bound together, and can only be moved together. The split can later be undone. When cloning, the two editors can be independently moved, closed, etc.

Putting two editors below each other can be especially useful when applying timing synchronization to the views of those editors.

By default the Eclipse IDE opens with one window, called the workbench window. Working with traces, various editors may be opened, typically on data that is related. When working with with multiple traces, especially when using multiple monitors, it can be of benefit to open a second workbench window. In this second workbench window, editors for a different trace can be opened, and analyzed. This keeps the editors/data for one trace in one window, separate from the editors/data of the other trace in the other window. The data for the different traces in the workbench windows on the different monitors can then for instance be compared. Also, timing synchronization is applied per workbench window. To open a new workbench window, click the Window  New Window menu.

Typically PPS works with large data sets. This complicates focussing on relevant data and the following functions therefore have been added to the Outline view.

This release prepares PPS to be open sourced.

The highlights of this release are:

The highlights of this release are:

This release of PPS adds the feature of defining, resolving and analyzing metric instances in a TMSC. When a metric instance exceeds its budget, a detailed root cause analysis can be performed to diagnose which task(s) caused this.

This first release of PPS includes a TMSC graphical viewer that is supported by a textual syntax.