Using Data Filters

This section introduces the data filter tool, or simply, the filter, and describes its use in the Trace Analyzer for filtering and searching operations.

Introduction

A filter is a user-configurable tool which is applied against an individual trace record to determine whether the data in that record meets a specific set of criteria.  A filter may represent a single condition or a compound condition.  A number of predefined data filters are shipped with the Trace Analyzer product.  These may be edited and new filters can be created through the interactive filter editing interface or the QuickFilter dialog window.  See also Filter Syntax.

For simplicity, the Trace Analyzer uses the same data filter tools both for search operations and for filter operations.  When used in these operations, a filter is applied against many trace records sequentially to find a specific record or to obtain a subset of records which all meet the filter's criteria.

Searching

Search jobs are used to find and select the next or previous trace record in the Overview Report which meets a specific filter's criteria.  To create a search job, a filter name is first selected from among the available filters using the Select Find Filter control in the Report Toolbar.  One of the Find Next or Find Previous icons is then clicked to submit the search job.  When the job is run, the search engine will begin testing the records contained in the current Overview Report against the selected filter.

The filter is applied to each trace record, starting at (but excluding) the currently selected record in the Overview Report and working forwards (Find Next) or backwards (Find Previous) from that record.  As soon as a record is encountered which meets the criteria specified in the filter, the search is terminated and that record is selected in the Overview Report table.  Progress of the search is reported via the job progress monitor in the Global Toolbar.

If a row in the Overview Report table is selected manually before the search completes, the search is abandoned, since it is assumed that the user's manual selection should take precedence over a previously launched search.  If the search is exhausted and no records have been found which meet the filter's criteria, the search is concluded without selecting a new record.  A search will not wrap from the end of the trace back to the beginning or from the beginning to the end.

Filtering

Filter jobs are used to refine the full set of records in a trace file (see Figure 1, for example) into a subset of records which all meet a specific filter's criteria (see Figure 2).  When a filter is applied to a trace, the set of records currently displayed in the Overview Report will be removed and replaced with this subset.  Note that if the filter's criteria cannot be matched to any records in the current trace, the Overview Report table will remain empty.

A filtering job is submitted by selecting the name of a filter from the Select Filter control in the Global Toolbar.  When the job is run, the filtering engine will test each record in the trace file against the selected filter.  Those records which meet the criteria of the filter, if any, are added to the new Overview Report.


Figure 1.  Unfiltered Overview Report


Figure 2.  Overview Report with SNA FID2 filter applied

Important Note:  Filters are not applied cumulatively.  A filtering job always applies its filter to the full set of trace data in the currently loaded trace file, even if the current Overview Report table contains records which already have been filtered.  If multiple filters should be applied to the data, they must be combined into a single filter expression using a conditional logical operator (see Filter Syntax).

To reset the Overview Report to the full, unfiltered data set, filtering must be turned off.  This is done by selecting the No Filter item in the Global Toolbar's Select Filter combo box.