[omniORB] Attribute Error In IDL Generated Module

[Duncan Grisby]

On Saturday 17 March, jiwils - Jimmy Wilson wrote:

[...]
> When importing the Test module below, I get an attribute error.  The
> attribute error's location is the module generated by omniidl for Test.IDL.

There is indeed a bug in omniidl (well, the Python back-end). Steffen
Ries correctly diagnosed the problem, but his fix breaks other
situations. I've checked in a full fix. The patch is small so it's at
the end. You can also get it from the normal channels -- CVS, nightly
snapshots, bugfixes patch.

As Steffen said, it's best to avoid doing typedef struct ... anyway,
since it's completely redundant in IDL.

Cheers,

Duncan.

--- src/lib/omniORBpy/omniidl_be/python.py	2000/11/01 11:25:56	1.27.2.4
+++ src/lib/omniORBpy/omniidl_be/python.py	2001/03/19 11:51:16
@@ -994,8 +994,19 @@
                    unaliased_type.kind() in [idltype.tk_struct,
                                              idltype.tk_union]:
 
-                    parent = dotName(fixupScopedName(unaliased_type.decl().\
-                                                     scopedName()))
+                    psname  = unaliased_type.decl().scopedName()
+                    myscope = decl.scopedName()[:-1]
+
+                    # If the struct/union definition is in the same
+                    # scope as the typedef, we must use a relative
+                    # name to refer to the parent class, since the
+                    # enclosing Python class has not yet been fully
+                    # defined.
+
+                    if psname[:len(myscope)] == myscope:
+                        parent = dotName(psname[len(myscope):])
+                    else:
+                        parent = dotName(fixupScopedName(psname))
 
                     self.st.out(typedef_struct_union_header,
                                 tdname = tdname,

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --