Trace Analysis Patterns (Part 25)

Here we introduce a software narratological partitioning of a trace into Head, Prologue, Core, Epilogue and Tail segments. This is useful for comparative software trace analysis. Suppose, a trace started just before the reproduction steps or a start marker was injected (by CDFMarker, for example) and finished just after the last repro steps or after an end marker was injected. Then its core trace messages are surrounded by prologue and epilogue statements. What is before and after are not necessary for analysis and usually distract an analyst. They are shown as grey areas on the following picture where the left trace is for working (non-working) scenario and the right trace is for non-working (working) scenario:

The size of a core segment need not be the same because environments and executed code paths might be different. However, often some traces are truncated (another pattern). Also sometimes it is difficult to establish whether the first trace is normal and the second has a tail or the first one is truncated and the second one is normal with an optional tail. Here artificial markers are important.

- Dmitry Vostokov @ DumpAnalysis.org + TraceAnalysis.org -

Leave a Reply