[omniORB] Exception not caught - omniORB-4.2.0

Thu Aug 6 15:13:00 UTC 2020

As I have said before, the changes in omniORB to support C++ 17's
removal of throw specifications are completely unrelated to the problem
you have on AIX.
The problem on AIX is that current versions of gcc generate incorrect
code. omniORB cannot and will not work around that.
The only known working compiler on AIX is IBM's xlc. You must either
switch to using xlc, or you must switch to an operating system on which
gcc works.
Duncan.

On Tue, 2020-08-04 at 04:14 +0000, Prasath_Palaniappan at amat.com wrote:
> Hi Duncan,
>  
> Sorry for bothering you.
>  
> As you are aware, we have been facing issue on using omniORB-4.2.0
> library in gcc-8 due to that exception is not caught properly after
> we upgraded from gcc-4.8 to gcc-8.2.
> Then we found the reason for the issue is gcc when using it on AIX.
> So, we raised this issue to gcc (please see the below mail) but we
> didn’t get any valid fix till now.
>  
> Meanwhile, we understand that the new released omniORB-4.2.4 version
> has fixes which is related to exception specifier to support latest
> C++17. 
> So, we believed the latest version might have been tested with gcc-
> 8.2 on AIX and we tried to check whether the exception handling issue
> (when we use with gcc-8.2 on AIX) is fixed within omniORB in the
> latest version.
> We compiled and checked the call_back examples and facing the same
> termination issue on throwing COMM_FAILURE exception. 
>  
> 
> Could you please let us know whether the latest omniORB-4.2.4
> provides support to gcc-8.2 on AIX? If not, what is the maximum
> supported and tested gcc version on AIX?
> Or do you have any plan for supporting that latest gcc version on
> AIX?
>  
> We are eagerly waiting for your response. Kindly suggest us to move
> forward in this issue.
>  
> Thanks & Regards,
> P. Prasath.
> _____________________________________________
> 
> 
> From: Prasath Palaniappan 
> 
> 
> Sent: Thursday, December 12, 2019 2:56 PM
> 
> 
> To: Duncan Grisby <duncan at grisby.org>; 
> omniorb-list at omniorb-support.com
> 
> 
> Cc: Shankar Chinnusamy <Shankar_Chinnusamy at amat.com>
> 
> 
> Subject: RE: [External] Re: [omniORB] Exception not caught - omniORB-
> 4.2.0
>  
>  
> Hi Duncan, 
>  
> We posted this query to gcc and got the response i.e. this is a bug
> in gcc on AIX. Please find the below link,
> https://www.spinics.net/lists/gcchelp/msg49970.html
>  
> Thank you so much for your prompt support.
>  
> Thanks & Regards,
> P. Prasath.
> _____________________________________________
> 
> 
> From: Prasath Palaniappan 
> 
> 
> Sent: Wednesday, December 11, 2019 7:32 PM
> 
> 
> To: 'Duncan Grisby' <duncan at grisby.org>; 
> omniorb-list at omniorb-support.com
> 
> 
> Cc: Shankar Chinnusamy <Shankar_Chinnusamy at amat.com>
> 
> 
> Subject: RE: [External] Re: [omniORB] Exception not caught - omniORB-
> 4.2.0
>  
>  
> Hi Duncan, 
>  
> As you suggested, I am looking into it on compiler and C++ runtime.
> I recompiled the gcc-8.2.0 with "--enable-threads" option in order to
> fix the issue as mentioned in gcc page 
> https://gcc.gnu.org/install/configure.html
>  
> I compiled omniORB and examples with this newly built gcc and tested
> the call_back example and ran into the same issue. 
> Gone through the traces and observed that omniDynamic library is not
> linked but this library does get linked and get initialized when we
> run the omniORB which is compiled with gcc-4.8.3.
> omniORB with gcc-4.8.3:
> omniORB: Initialising omniDynamic library.
> omniORB with gcc-8.2.0:
> omniORB: (0) 2019-12-11 01:33:35.484672: Information: the omniDynamic
> library is not linked.
>  
> I agree this is an information, but curious to know whether it
> impacts anything since it is named as dynamic library.
> I looked into forum and found the same post in below link,
> https://omniorb-list.realvnc.narkive.com/3WcwVC6I/omniorb-information-the-omnidynamic-library-is-not-linked
> and tried your solution by setting "export ORBendPointPublish=ipv4". 
> It gets reflected in configuration as below but still the omniDynamic
> is not linked.
> omniORB:   endPointPublish = ipv6
> And also another query in below link which seems the same, but not
> answered yet.
> https://www.omniorb-support.com/pipermail/omniorb-list/2010-February/030534.html
>  
> I am not blaming omni code. I am looking forward to you throwing some
> light to proceed further in this issue.
>  
> Thanks & Regards,
> P. Prasath.
>  
> -----Original Message-----
> 
> 
> From: Duncan Grisby <duncan at grisby.org>
> 
> 
> 
> Sent: Wednesday, November 6, 2019 9:53 PM
> 
> 
> To: Prasath Palaniappan <Prasath_Palaniappan at amat.com>; 
> omniorb-list at omniorb-support.com
> 
> 
> Cc: Shankar Chinnusamy <Shankar_Chinnusamy at amat.com>
> 
> 
> Subject: [External] Re: [omniORB] Exception not caught - omniORB-
> 4.2.0
>  
> CAUTION: EXTERNAL EMAIL. Verify before you click links or open
> attachments. Questions? Contact GIS.
>  
>  
> On Tue, 2019-11-05 at 18:06 +0000, Prasath via omniORB-list wrote:
>  
> [...]
> > But when we run the omniORB + application which is compiled with
> gcc- 
> > 8.2.0, omniORB gets crashed because it unable to catch the
> Exception 
> > “CommFailure” and lead into termination.
>  
> This is exactly the same thing you reported back in April:
>  
> https://www.omniorb-support.com/pipermail/omniorb-list/2019-April/032050.html
>  
> The answer now is the same as the answer then. omniORB's code is
> correct. It has code that catches the exception. For some reason in
> your system it is not being caught, but that is not omniORB's fault.
>  
> [...]
> > But the exception does not get caught with the catch when compiled 
> > omniORB with gcc-8.2.0.
> > Could you please clarify us the cause for this issue and why
> ominORB 
> > unable to handle the exception if gcc-8.2.0 and also provide your 
> > suggestions to fix this issue?
>  
> Why do you insist on thinking that omniORB is to blame here?  As you
> say yourself, the same code works totally fine when you use gcc
> 4.8.3.
> The one thing you have changed is the compiler version. THAT is the
> culprit, not omniORB. My suggestion to fix the issue is to find a
> compiler that works.
>  
> Duncan.
>  
> --
>  -- Duncan Grisby         --
>   -- duncan at grisby.org     --
>    -- http://www.grisby.org --
>  
>  
> The content of this message is APPLIED MATERIALS CONFIDENTIAL. If you
> are not the intended recipient, please notify me, delete this email
> and do not use or distribute this email. 
> 
> 
> 
-- 
 -- Duncan Grisby         --
  -- duncan at grisby.org     --
   -- http://www.grisby.org --

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.omniorb-support.com/pipermail/omniorb-list/attachments/20200806/cdc62734/attachment.html>