Merged revisions 60210-60233 via svnmerge from

svn+ssh://pythondev@svn.python.org/python/trunk

........
  r60213 | christian.heimes | 2008-01-23 15:00:25 +0100 (Wed, 23 Jan 2008) | 1 line

  Use Py_TYPE() instead of ->ob_type
........
  r60214 | armin.rigo | 2008-01-23 15:07:13 +0100 (Wed, 23 Jan 2008) | 3 lines

  patch 1754489 by vlahan:
  improve portability of address length calculation for AF_UNIX sockets
........
  r60216 | christian.heimes | 2008-01-23 15:20:50 +0100 (Wed, 23 Jan 2008) | 1 line

  Fixed bug #1915: Python compiles with --enable-unicode=no again. However several extension methods and modules do not work without unicode support.
........
  r60221 | christian.heimes | 2008-01-23 18:15:06 +0100 (Wed, 23 Jan 2008) | 2 lines

  Applied #1069410
  The "can't load dll" message box on Windows is suppressed while an extension is loaded by calling SetErrorMode in dynload_win.c. The error is still reported properly.
........
  r60224 | guido.van.rossum | 2008-01-23 21:19:01 +0100 (Wed, 23 Jan 2008) | 2 lines

  Fix two crashers.
........
  r60225 | kurt.kaiser | 2008-01-23 23:19:23 +0100 (Wed, 23 Jan 2008) | 3 lines

  Could not open files in .idlerc directory if latter was hidden on Windows.
  Issue 1743, Issue 1862.
........
  r60226 | guido.van.rossum | 2008-01-23 23:43:27 +0100 (Wed, 23 Jan 2008) | 2 lines

  Fix misleading comment reported in issue #1917.
........
  r60227 | kurt.kaiser | 2008-01-23 23:55:26 +0100 (Wed, 23 Jan 2008) | 2 lines

  There was an error on exit if no sys.exitfunc was defined. Issue 1647.
........
  r60228 | guido.van.rossum | 2008-01-24 00:23:43 +0100 (Thu, 24 Jan 2008) | 2 lines

  Turn three recently fixed crashers into regular tests.
........
  r60229 | raymond.hettinger | 2008-01-24 01:54:21 +0100 (Thu, 24 Jan 2008) | 1 line

  Add first-cut at an approximation function (still needs rounding tweaks).  Add continued fraction conversions.
........
  r60230 | raymond.hettinger | 2008-01-24 03:00:25 +0100 (Thu, 24 Jan 2008) | 1 line

  Minor clean-up and more tests.
........
  r60231 | raymond.hettinger | 2008-01-24 03:05:06 +0100 (Thu, 24 Jan 2008) | 1 line

  Cleanup
........
  r60232 | neal.norwitz | 2008-01-24 05:14:50 +0100 (Thu, 24 Jan 2008) | 1 line

  Fix the tests by restoring __import__.  I think the test is still valid.
........
  r60233 | neal.norwitz | 2008-01-24 08:40:51 +0100 (Thu, 24 Jan 2008) | 4 lines

  Fix the test_urllib2net failures that were caused by r58067.
  I'm not sure this is the correct fix, but at least the test passes
  now and should be closer to correct.
........

1 files changed, 9 insertions, 0 deletions

diff --git a/Python/dynload_win.c b/Python/dynload_win.c
index 39c091b..4db12c4 100644
--- a/Python/dynload_win.c
+++ b/Python/dynload_win.c
@@ -171,11 +171,16 @@ dl_funcptr _PyImport_GetDynLoadFunc(const char *fqname, const char *shortname,
 		HINSTANCE hDLL = NULL;
 		char pathbuf[260];
 		LPTSTR dummy;
+		unsigned int old_mode;
 		/* We use LoadLibraryEx so Windows looks for dependent DLLs 
 		    in directory of pathname first.  However, Windows95
 		    can sometimes not work correctly unless the absolute
 		    path is used.  If GetFullPathName() fails, the LoadLibrary
 		    will certainly fail too, so use its error code */
+
+		/* Don't display a message box when Python can't load a DLL */
+		old_mode = SetErrorMode(SEM_FAILCRITICALERRORS);
+
 		if (GetFullPathName(pathname,
 				    sizeof(pathbuf),
 				    pathbuf,
@@ -183,6 +188,10 @@ dl_funcptr _PyImport_GetDynLoadFunc(const char *fqname, const char *shortname,
 			/* XXX This call doesn't exist in Windows CE */
 			hDLL = LoadLibraryEx(pathname, NULL,
 					     LOAD_WITH_ALTERED_SEARCH_PATH);
+
+		/* restore old error mode settings */
+		SetErrorMode(old_mode);
+
 		if (hDLL==NULL){
 			PyObject *message;
 			unsigned int errorCode;