Abstract: A method for generating trace information for an exception in a software application, the software application having a traced mode for the generation of trace information and an untraced mode, and the software application further having a retry point as a defined point in a series of instructions for the execution of the software application, the method comprising the steps of: commencing execution of the software application in the untraced mode; on execution of the retry point, recording execution state information for the retry point; on occurrence of an exception at an exception point in the software application, reverting execution to the retry point using the recorded execution state information and switching to the traced mode; and in response to a determination that the exception point is executed in the traced mode, causing execution of the software application to proceed in the untraced mode.

Abstract: The present invention discloses a solution that defines multiple retry points for a software application which are operable to generate and record trace information only as part of a retry execution in response to an exception. Outside these conditions, the burdens of generating and recording trace information can be avoided. Furthermore, the determination of a preferred one of these retry points can be made to focus the retry execution in a way that only the retry point which generates the most helpful, or informative, trace information is necessarily reverted to in response to an exception.

Abstract: Trace information is selectively generated for a software routine based on the perceived reliability of the software routine. The software routine includes at least one trace point having an active state and an inactive state. A previously-established reliability indicator for the software routine is read before the routine is executed. The reliability indicator is based on criteria such as age, prior level of testing, source, number or previously detected faults and/or number of prior successful executions. If the reliability indicator meets a predetermined threshold, the active state is selected for the trace point. If the reliability indicator does not meet the predetermined threshold, the inactive state is selected for the trace point.

Abstract: A method for generating trace information for an exception in a software application, the software application having a traced mode for the generation of trace information and an untraced mode, and the software application further having a retry point as a defined point in a series of instructions for the execution of the software application, the method comprising the steps of: commencing execution of the software application in the untraced mode; on execution of the retry point, recording execution state information for the retry point; on occurrence of an exception at an exception point in the software application, reverting execution to the retry point using the recorded execution state information and switching to the traced mode; and in response to a determination that the exception point is executed in the traced mode, causing execution of the software application to proceed in the untraced mode.

Abstract: The present invention discloses a solution that defines multiple retry points for a software application which are operable to generate and record trace information only as part of a retry execution in response to an exception. Outside these conditions, the burdens of generating and recording trace information can be avoided. Furthermore, the determination of a preferred one of these retry points can be made to focus the retry execution in a way that only the retry point which generates the most helpful, or informative, trace information is necessarily reverted to in response to an exception.