Open debugger at the location of failing assert, if possible.

[wxWidgets.git] / tests / runtests.bat

diff --git a/tests/runtests.bat b/tests/runtests.bat
index 63521ff771247aadd68167f75af351813dd69637..08362a5b2ed20770d6dd8dab723cde131e926c64 100644 (file)
--- a/tests/runtests.bat
+++ b/tests/runtests.bat
@@ -3,29 +3,66 @@
 REM Runs wxWidgets CppUnit tests
 REM This script is used to return the correct return value to the caller
 REM which is required by Buildbot to recognize failures.
+REM Note that in DOS error level is not the return code of the previous
+REM command; it is for (some!) built-in DOS commands like FIND but
+REM in general it's not. Thus to get the return code of the test utility
+REM we need some hack; see the guide:
+REM    http://www.infionline.net/~wtnewton/batch/batguide.html
+REM for general info about DOS batch files.
+
+REM Author: Francesco Montorsi
 REM $Id$
 
 rem set the path for running the tests if they use DLL build of wx
-for /d %%x in (..\lib\*_dll) do @PATH=%x;%PATH%
+for /d %%x in ("..\lib\*_dll") do @set PATH=%%x;%PATH%
 
 set failure=0
 
 for /d %%x in (*) do @(
     if exist %%x\test.exe (
-        %%x\test.exe
-        if %errorlevel% GEQ 1 set failure=1
+        %%x\test.exe >tmp
+        
+        REM show the output of the test in the buildbot log:
+        type tmp
+        
+        REM hack to understand if the tests succeeded or not
+        REM (failure=1 is set if "OK" does not appear in the test output)
+        type tmp | find "OK" >NUL
+        if ERRORLEVEL 1 set failure=1
+
+        REM separe the output of the test we just executed from the next one
+        echo.
+        echo.
+        echo ========================================================================
+        echo ========================================================================
+        echo.
+        echo.
     )
+    
     if exist %%x\test_gui.exe (
-        %%x\test_gui.exe
-        if %errorlevel% GEQ 1 set failure=1
+        %%x\test_gui.exe >tmp
+        
+        REM show the output of the test in the buildbot log:
+        type tmp
+        
+        REM hack to understand if the tests succeeded or not
+        REM (failure=1 is set if "OK" does not appear in the test output)
+        type tmp | find "OK" >NUL
+        if ERRORLEVEL 1 set failure=1
     )
 )
 
 REM exit with code 1 if any of the test failed
-if %failure% EQU 1 exit /b 1
+del tmp
+if %failure% EQU 1 (
+    echo.
+    echo One or more test failed
+    echo.
+    exit 1
+)
 
 REM remove the failure env var:
 set failure=
 
 REM exit with code 0 (all tests passed successfully)
-exit /b 0
+exit 0