Bug in particle editor (with fix) (actually bug in SimBase)

I don't get it. Why are there no unit tests for this code? It claims to have been around for almost ten years, and thus be "mature".

The built-in particle editor writes out scripts that you can load in as particle data. However, if you do, you actually get a crash, because it turns out that it creates data blocks with "new" rather than "datablock," because, in turn, the serialization support in SimBase is not smart enough to tell the difference.

I'm sorry if I seem ornery, but this is quite frustrating. Each time I try to use something, it turns out to be broken and un-tested. Last week, it was the Linux build; this week, it was object serialization AND cross-platform communication. I'd really like to be able to actually write some content code, rather than debug the core product...

Fix:

===================================================================
--- console/simBase.cc  (revision 648)
+++ console/simBase.cc  (working copy)
@@ -446,6 +446,14 @@
       mFieldDictionary->writeFields(this, stream, tabStop);
 }

+static char const * getNewFunction(SimObject * so)
+{
+  if (dynamic_cast<SimDataBlock *>(so) != NULL) {
+    return "datablock";
+  }
+  return "new";
+}
+
 void SimObject::write(Stream &stream, U32 tabStop, U32 flags)
 {
    // Only output selected objects if they want that.
@@ -454,7 +462,8 @@

    writeTabs(stream, tabStop);
    char buffer[1024];
-   dSprintf(buffer, sizeof(buffer), "new %s(%s) {\r\n", getClassName(), getName() ? getName() : "");
+   dSprintf(buffer, sizeof(buffer), "%s %s(%s) {\r\n", getNewFunction(this),
+      getClassName(), getName() ? getName() : "");
    stream.write(dStrlen(buffer), buffer);
    writeFields(stream, tabStop + 1);
    writeTabs(stream, tabStop);

By the way: the default key for the particle editor (F9) needs to change, given that the debug render mode uses F9.