I noticed inconsistency in observations report across formats - XML only shows the first stack while TXT shows all instances of any error. For example, there is an error P36 - showing 48 instances in observations.txt, but only shows one in observations.xml
"C:\Program Files (x86)\Intel\Inspector XE\bin32\inspxe-cl.exe" -report observations -report-all -format xml -report-output intelinspector_results\report_mi2_000_xml\intelinspector_observations.xml -result-dir intelinspector_results/mi2_000
Intel Inspector XE 2013 Update 8
Is this a known issue?
I think what you are seeing is that in the .txt report the observations are being broken out into all the individual call stacks that make them up and in the .xml file they are not being displayed that way. Basically you can get to the same place in the code through several different code paths, but if the results are the same they are the same observation.
The report designer says that the data that goes into report-all for .txt and .xml files is the same, but it is possible that the schema that is displaying the .xml is not presenting all the information. If you want to attach the reports (or snippets of them) or email me at firstname.lastname@example.org, he would be willing to take a look at them.
Here is the final outcome, from the designer:
After further analysis, it turns out that the differences in the report between xml and text formats is due to the fact that xml reports can only be created in hierarchical format. In other words, whether requesting an observations report or a problems report, when the format specified is xml, the report is a problems report. I am going to add an error message to the product indicating that an xml report can only be generated in hierarchical format which will be printed to the screen when the user requests an observations type report and an xml format.
In observations reports, each observation is individually reported even if duplicates of the same observation exist. Usually, the difference is the stack or path leading to the same observation.
In problem reports, observations are grouped together and reported in hierarchical format which matches what the GUI displays. Observations are grouped in problem sets. Each problem set is reported at the top level, then each problem in the problem set is reported in the subsequent level, then each observation of the problem is reported in the subsequent level. However, observation duplicates are eliminated. If the more detailed report is requested using –report-all, duplicates are still eliminated but their stacks are reported to indicate the many paths leading to the same diagnostic.