-
The GCaldb::filename method returns an absolute IRF filename based on a set of selection criteria. The selection criteria are still very simple, and a more sophisticated should be implemented in the futur. For the time being, the method does the job.
-
The results of the Python unit test were not correctly tracked due to some subtle problems related to the way how C++ code is called through the Python interface. A kluge has been found and added to GTestSuite.i to retrieve the Python test results and to propagate them to the C++ test suite, so now Python test results are also written into the XML file, and the Python tests signal failure in case that something goes wrong.
-
GTestSuite::run starts with creating a test case that collects the total time spent in a test and that also collects any failures of the individual test cases. Before, this test case was appended AFTER running the individual test cases, which led to an XML file out of order. Now the general test case is appended BEFORE running the test cases.
-
The vo module and GVOClient class were missing in the Python interface. They have been added now.
-
This was the initial intention, but the code did the opposite.
-
GAMMALIB_CALDB should take precedence over CALDB, yet so far the opposite was true. This has now been corrected for.
-
An issue has been raied (#1098) to track solving the problem.
-
The attribute for identifiers is "id". This has been corrected in GCTAModelBackground. Further, writing of the attribute has been added to GModelSky. Attributes are also only written if the identifier string is not empty.
-
The testconf.sh.in script may be used to test the GammaLib configuration script following installation. This script should be used by the continous integration system to guarantee the coherence of the installation script at any time.
-
The saveto() method now also checks whether a file with the .gz suffix exists.
-
The former trigonometric method led to different results on different platforms due to numerical imprecisions. We now use Huilier's theorem to be less sensitive to numerical problems. The polygon is decomposed in two triangles for this purpose.
-
Conflicts: ChangeLog
-
Conflicts: src/sky/GWcs.cpp
-
When a pixel touches the pole, the pixel is effectively a triangle and not a polygon. This case is now handled properly.
