Bug 845 - IDL compiler generates redundant code for AMI
Summary: IDL compiler generates redundant code for AMI
Status: RESOLVED FIXED
Alias: None
Product: TAO
Classification: Unclassified
Component: IDL Compiler (show other bugs)
Version: 1.4.4
Hardware: All All
: P3 normal
Assignee: Johnny Willemsen
URL:
Depends on: 1369 1917
Blocks:
  Show dependency tree
 
Reported: 2001-04-01 21:50 CDT by Carlos O'Ryan
Modified: 2006-08-10 07:57 CDT (History)
0 users

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Carlos O'Ryan 2001-04-01 21:50:55 CDT 
For each ExceptionHolder class multiple raise_XXX() methods are generated, all
these methods are identical except for the list of user exceptions that are
acceptable.  However, the list of user exceptions is easily represented as an
array, in fact the code *uses* an array.

This screams for a subroutine, in fact, the code is almost an exact duplicate of
what TAO_GIOP_Twoway_Invocation::invoke() does!  All this code should be
factored into a few helper functions and invoked from both the Invocation and
the generated code.
Comment 1 Carlos O'Ryan 2001-04-02 10:27:10 CDT 
Fixed typo in summary
Comment 2 DOC Center Support List (internal) 2001-04-10 12:22:53 CDT 
Accepted.
Comment 3 Carlos O'Ryan 2004-11-19 15:15:41 CST 
Seems to me like the techniques described in bug 1369 can be used to fix this
bug too.
Comment 4 Johnny Willemsen 2005-03-24 06:32:22 CST 
When I look at the generated code from tests/AMI I just see the following being 
generated for each raise method. Seems to be fixed or not?

  TAO_Messaging_Helper::exception_holder_raise (
    0,
    0,
    this->marshaled_exception ().get_buffer (),
    this->marshaled_exception ().length (),
    this->byte_order (),
    this->is_system_exception ()
    ACE_ENV_ARG_PARAMETER);
Comment 5 Carlos O'Ryan 2005-03-24 06:43:55 CST 
Yes, the AMI code is using arrays now.  Do notice that the regular synchronous
requests and the AMI holders need the same arrays, but the arrays are not shared
between the two routines.  IMHO there is opportunity to gain a small footprint
reduction in that code.
Comment 6 Johnny Willemsen 2005-03-24 06:46:48 CST 
Updated version to 1.4.4, still a little issue that we can improve.
Comment 7 Johnny Willemsen 2005-08-07 13:05:05 CDT 
AMI exception handling is changed in the latest corba spec, added depends on bug
1917. When that one is fixed I think this is then invalid.
Comment 8 Johnny Willemsen 2005-10-19 03:23:50 CDT 
taking over this bug, I am working on the new AMI mapping
Comment 9 Johnny Willemsen 2005-11-02 03:16:25 CST 
accept
Comment 10 Johnny Willemsen 2006-08-09 08:21:26 CDT 
Carlos, do you maybe have some time to recheck this with 1.5.2 which has the new
AMI mapping support
Comment 11 Carlos O'Ryan 2006-08-10 07:57:32 CDT 
Johnny is right.  The code does not have any of the duplications originally
documented in this bug.