omniORB 3.0.3 Bug List

attlogo6876_dk.gif (2280 bytes)

sideHome.gif (2321 bytes)sideDownload.gif (2450 bytes)sideDocumentation.gif (2512 bytes)sideFAQ.gif (2344 bytes)sidePatch.gif (2543 bytes)sideSearch.gif (2403 bytes)

The following bugs in omniORB 3.0.3 have been fixed. You can get the fixes in three ways:

  • Update from CVS in the "omni3_develop" branch.
  • Apply this patch to the omniORB 3.0.3 distribution.
  • Download the latest source snapshot. (Note that this is generated nightly, so the latest bugfixes may not appear until tomorrow.)

The bugs pages for earlier versions can be found here:


Summary: System clock change causes the scavenger thread to consume lots of CPU cycles (bug number 7)
Date: Tue May 29 12:14:54 BST 2001
Fixed by: sll
Reported by: Peter Rönnquist
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-05/0195.html
Description: Scavenger now get the real time after each scan. This is to cope with the system clock set backward by a large amount. Seems to happen a lot to some system..

Summary: Incorrect namespaces for MSVC++ constant work-around (bug number 6)
Date: Fri Apr 27 11:58:25 BST 2001
Fixed by: dpg1
Reported by: Daniel Bell
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-03/0107.html
Description: The code generated to give constants external linkage on MSVC failed for nested modules.

Summary: Incorrect code generated for includes at non-file scope (bug number 5)
Date: Wed Apr 25 17:49:17 BST 2001
Fixed by: dpg1
Reported by: Olaf Meding
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-04/0152.html
Description: The IDL compiler C++ back-end would generate incorrect code for #includes at non-file scope. Now, declarations inside files #included at non-file scope are treated as if they appeared in the including file.

Summary: Bug in ORB core causes assertion failure (bug number 4)
Date: Tue Mar 27 17:56:31 BST 2001
Fixed by: djr
Reported by: Roumen Ivanov
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-03/0159.html
Description: A bug in the ORB core would causes an assertion failure when an object is deactivated. The problem occurs if the object has local references whose c++ type disagrees with that of the object.

This may explain a problem with omniNames, given in the link above.

Summary: Incorrect TypeCode for union with multiple case labels (bug number 3)
Date: Wed Mar 21 16:31:13 GMT 2001
Fixed by: dpg1
Reported by: Clemens Fischer
Description: The generated TypeCode for a union with multiple case labels would have the wrong member count. e.g. the count for

  union U switch (long) {
    case 1:
    case 2: long l;
    case 3: string s;
  };
  
would be 2 instead of 3.

Summary: Incorrect DynSK stubs for some recursive types (bug number 2)
Date: Tue Mar 20 16:55:07 GMT 2001
Fixed by: dpg1
Reported by: Lars Immisch
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-03/0101.html
Description: The IDL compiler C++ back-end would generate incorrect TypeCode constants for IDL like

    struct S {
      sequence <sequence <S> > a;
    };
  

Summary: Memory corruption when initialising multiply-recursive structures (bug number 1)
Date: Tue Mar 20 16:55:07 GMT 2001
Fixed by: dpg1
Reported by: Marcus Bullingham
Link for this bug: http://www.uk.research.att.com/omniORB/archives/2001-02/0259.html
Description: IDL types where there was more than one recursion to the same type would corrupt memory, sometimes leading to a crash.


For comments, feedback, etc, please see the 'Keeping in touch' page.
Copyright 2000 - AT&T Laboratories Cambridge