headers
Karel Gardas | 2 Sep 2006 09:00
Favicon

Re: No time parameter on ORB::perform_work()?


Donald,

On Thu, 31 Aug 2006, Donald A. Dade wrote:

> Hello Sorin,
>
> Thanks for your reply.
>
> I intended the question as a MICO related one. I'm aware of all the
> systems programming techniques, at least the POSIX ones, don't really
> feel good about any of them because they are all hacks by definition,
> and was really asking if I was missing something made available by CORBA
> itself like the TAO's ORB::perform_work() which accepts a time interval
> a la libc's pselect().

I'd like to correct you a bit here. First of all, CORBA itself does not 
define any parameter for ORB::perform_work() operation. If it is added by 
TAO, then it is just proprietary extension. If you need something like 
that, we in MICO rather prefer as clean as possible CORBA API implemented 
together with command-line arguments which "tune" some behaviour. In your 
case, I would propose to add something like -ORBPerformWorkTimeout 
<number> command-line parameter.

> I've hardly looked at the source, but wouldn't it be pretty easy for me
> to implement it? If I recall, doesn't ORB::run() just call pselect() and
> wait indefinitely? So a time-bounded ORB::perform_work() could be the
> same call except with the struct timeval passed in?
>
> In order to contribute, do I have to make it work for win32 also?
(Continue reading)

Permalink | Reply | 0 comments |
headers
Karel Gardas | 2 Sep 2006 09:14
Favicon

Re: Mico on mac (intel platfrom)


Hi,

I'm afraid that MacOS X/ia32 platform is quite new (considering nobody in 
MICO community using Darwin for Intel) that nobody probably will be able 
to help you from this issue. That indeed might be anything, but have you 
do the same test on Linux/ia32 and MacOSX/ppc platforms? Did it work well? 
If so, then I would try to debug more hard the test case. It seems it 
might be race-condition somewhere either in Mico or in MacOSX/ia32, since 
what you see is basically wait for incomming reply. Please also use 
tcpdump to check if your server already received the reply and if so, then 
you know something is definitelly broken.

Good luck!
Karel
--
Karel Gardas                  kgardas <at> objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com
---
Need experienced, fast, reliable technical MICO support?
---> http://www.objectsecurity.com/mico_commsup_referral.html <---
---

On Fri, 25 Aug 2006, Gurminder Singh wrote:

> Hi, all
>
> I am really stuck in the following issue.
>
> I have compiled the MICO (2.3.12 ) on mac (10.4.6)  for i386
(Continue reading)

Permalink | Reply | 501 comments |
headers
Karel Gardas | 2 Sep 2006 09:16
Favicon

Re: mico 2.3.12 and Visual C++ v8

On Thu, 24 Aug 2006, J. Scott Evans wrote:

>
> Hello,
>
> I downloaded the package mico-cl8-15.05.06.zip and successfully compiled
> it with VS 2005.  I'm also able to successfully run micod.  However, when
> I run an imr command, even just the imr list command, I get a popup window
> with the following information:
>
>
> Microsoft Visual C++ Debug Library
>
> Debug Assertion Failed!
>
> Program: C:\Evans\..\mico\bin\micod.exe
> File: C:\Program Files\Microsoft Visual Studio 8\VC\INCLUDE\vector
> Line: 756
>
> Expression: vector subscript out of range
>
> For information on how your program can cause an assertion
> failure, see the Visual C++ documentation on asserts.
>
> (Press Retry to debug the application)
>
>
> Any idea why the imr/micod is causing this error.  I do not get the error
> with VS 2003, the imr works as expected.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Karel Gardas | 2 Sep 2006 09:18
Favicon

Re: mico/cygwin


On Thu, 24 Aug 2006, J. Scott Evans wrote:

>
> Hello,
>
> I experienced the same problem as described below regarding missing ssl
> lib during linking.  I hacked the Makefile so it includes the ssl and
> crypto libs and the link problem was resolved.
>
> However, I then encountered another problem that mystifies me.  Below is
> the error message.  Any ideas?  BTW, I'm using gcc 3.3.3 under cygwin.
>
> Thanks,
> Scott
>
> make[1]: Entering directory `/cygdrive/c/Downloads/Software/MICO/mico/coss'
> for i in  naming events property time; do make -C $i idlcompile || exit 1;
> done
> make[2]: Entering directory
> `/cygdrive/c/Downloads/Software/MICO/mico/coss/namin
> g'
> /cygdrive/c/Downloads/Software/MICO/mico/./idl/idl -B../.. --c++-skel
> --name Cos
> Naming \
>        ../../include/coss/CosNaming.idl
> [1090640593 - 0x100e0020] os-thread/pthreads.cc:343: assertion failed
> Signal 6
> make[2]: *** [CosNaming.cc] Error 134
> make[2]: Leaving directory
(Continue reading)

Permalink | Reply | 501 comments |
headers
Karel Gardas | 2 Sep 2006 09:20
Favicon

Re: universal binary for mico


Hi,

On Thu, 24 Aug 2006, Gurminder Singh wrote:

> Hi, all
>
> Has any one compiled mico on  Mac  for intel (i386)  and PPC
> architecture both (universal binary).

if you do any changes to MICO for your Mac/ia32 architecture, please also 
submit them for review and for inclusion. Your platform is quite new and 
might be also interesting for other MICO users.

Cheers,
Karel
--
Karel Gardas                  kgardas <at> objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com
Permalink | Reply | 0 comments |
headers
Karel Gardas | 2 Sep 2006 09:23
Favicon

Re: Mico Generated Code Question


Hello Paul,

have you tried POA tie approach? IIRC it was designed to fulfil such 
needs, i.e. integrating legacy code.

Cheers,
Karel
--
Karel Gardas                  kgardas <at> objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com
---
Need experienced, fast, reliable technical MICO support?
---> http://www.objectsecurity.com/mico_commsup_referral.html <---
---

On Thu, 24 Aug 2006, Caswell, Paul (SELEX Comms) (UK) wrote:

>
>
> Hello All,
> Looking at the code generated from IDL I was wondering why none of the
> base classes are pure C++ interfaces.  For example, given the following
> IDL:
>
> interface Base
> {
>   void foo();
>   void bar();
> };
(Continue reading)

Permalink | Reply | 0 comments |
headers
Siemens Eduard | 5 Sep 2006 10:24

Compiling mico w/o ssl support

Hello folks, 
I'm using mico in some distributed applications. The main issue is the
dependancy on the libssl.so
Since different machines runs on different Llinux distributions, we
can't just distribute binaries to all the machines because of 
Different versions of used openssl. However, we use those applications
only in our corporate environment, so the ssl-support within mico is not
really necessary. But I can't find a way to configure mico without
ssl-support. I tried ./configure --without-ssl, but it still link ssl
routines to the code.

Best regards, 
Eduard Siemens

========================================================================
==
Dr. Eduard Siemens                    mailto:eduard.siemens <at> thomson.net
Senior Development Engineer           http://www.thomson.net
Hannover Network Protocols Lab (HNP)       phone : +49 511 418 2168
Corporate Research Hannover           fax   : +49 511 418 2483
THOMSON / Karl-Wiechert-Allee 74 / D-30625 Hannover
Permalink | Reply | 0 comments |
headers
Jon Azpiazu | 6 Sep 2006 08:13
Picon

error in template.h

Hello,

Sometimes (not always) when I try to shutdown a consumer, I get the
following error:

.../mico-2.3.12//include/mico/template.h:102:
assertion failed Aborted

This happens when I try to exit cleanly after receiving a Ctrl + C

orb->shutdown (FALSE);
orb->destroy ();

Any idea?

Thanks
Permalink | Reply | 1 comment |
headers
Jon Azpiazu | 6 Sep 2006 08:19
Picon

notified about disconnections

Hello,

When running a supplier and several consumers, I need that my
consumers get notified when I shutdown the supplier. Otherwise, they
would just get stalled, waiting for a push().
I thought the solution would be calling to disconnect_push_consumer()
on the proxy_consumer , but it does not seem to have any effect.
What is the proper way of doing this?

Thank you
Permalink | Reply | 2 comments |
headers
Jon Azpiazu | 6 Sep 2006 09:21
Picon

Re: notified about disconnections

Ok, I think I can answer myself.
The solution that seems to work for me, is calling to
event_channel->destroy() in the Supplier, which makes a call to
disconnect_push_consumer() in the Consumers.

On 9/6/06, Jon Azpiazu <jon.azpiazu <at> gmail.com> wrote:
> Hello,
>
> When running a supplier and several consumers, I need that my
> consumers get notified when I shutdown the supplier. Otherwise, they
> would just get stalled, waiting for a push().
> I thought the solution would be calling to disconnect_push_consumer()
> on the proxy_consumer , but it does not seem to have any effect.
> What is the proper way of doing this?
>
> Thank you
>
Permalink | Reply | 1 comment |

Gmane