Thread: [Orbit-python-list] ORBit - MICO
Status: Inactive
Brought to you by:
tack
From: Panagiotis I. <tak...@sk...> - 2002-11-22 15:18:31
|
Hi, I'm trying to set up a CORBA Component system. I've searched for Open Source ORB's implementing at least a part of these standards and stumbled upon MICO or actually MicoCCM. Luckily the clientside of a CORBA system using components doesn't have to know about CORBA components at all, so you can use any CORBA 2.3 (or 2.4?) compliant ORB to use the components services. Because I like Python a lot, I wanted to use Python CORBA bindings and because I've played around with Orbit-Python before, I decided to try this one first. Half a year ago, I had lots of trouble getting _any_ ORB to communicate with ORBit (omniORB, TAO, MICO). At this time, I still didn't manage to get ORBit 0.x communicating with other ORBs but I'm getting some good results with ORBit 2.5.0. I'm using Debian GNU/Linux with orbit-python 1.99.0-4 on Python 2.2 with ORBit version 2.5.0. I'm using MICO 2.3.7. The MICO C++ server works, and also works with a MICO C++ client. The ORBit Python client works partly. When using a MICO server with an ORBit Python client, the operations get invoked on the CORBA object (server), but the client application aborts showing the following message: ** ERROR **: file orbit-object.c: line 146 (do_unref): assertion failed: (robj->refs < ORBIT_REFCOUNT_MAX && robj->refs > 0) On the MICO serverside: pissaris@nyx:~/robot/cxx/a/b$ g++ main.o orocos.o orocos_impl.o -o blaai -lmico2.3.7 -ldl orocos_skel.o pissaris@nyx:~/robot/cxx/a/b$ ./blaai found orb found boa a Server reference: IOR:010000001700000049444c3a4f726f636f732f4b756b613336313a312e300000020000000000000028000000010100000c0000003139322e3136382e302e3100ee8600000c000000424f41c0a8000100006789010100000024000000010000000100000001000000140000000100000001000100000000000901010000000000 echo STARTING KUKA STOPPING KUKA STARTING KUKA STOPPING KUKA As you can see, multiple invocations passed through (every invocation causes one outputline of "STARTING KUKA","STOPPING KUKA"). The serverside works just fine. On the ORBit Python clientside: pissaris@nyx:~/robot/cxx/a/b$ python takis.py 1 <Orocos.Kuka361 instance at 0x816888c> Invoking operation on CORBA server Ending CORBA client ** ERROR **: file orbit-object.c: line 146 (do_unref): assertion failed: (robj->refs < ORBIT_REFCOUNT_MAX && robj->refs > 0) aborting... Afgebroken pissaris@nyx:~/robot/cxx/a/b$ python takis.py 1 <Orocos.Kuka361 instance at 0x816888c> Invoking operation on CORBA server Ending CORBA client ** ERROR **: file orbit-object.c: line 146 (do_unref): assertion failed: (robj->refs < ORBIT_REFCOUNT_MAX && robj->refs > 0) aborting... Afgebroken Not directly related to the above: When I use ORBit Python by itself (no MICO or anything else involved), using void as return type causes an immediate segfault when using the Python client: pissaris@nyx:~/robot/cxx/a$ ./server found orb found boa a Server reference: IOR:010000001700000049444c3a4f726f636f732f4b756b613336313a312e300000020000000000000028000000010100000c0000003139322e3136382e302e3100248600000c000000424f41c0a800010000218b010100000024000000010000000100000001000000140000000100000001000100000000000901010000000000 STARTING KUKA STARTING KUKA STARTING KUKA STARTING KUKA STARTING KUKA pissaris@nyx:~/robot/cxx/a$ python takis.py 1 <Orocos.Kuka361 instance at 0x8166864> Invoking operation on CORBA server Segmentatie fout When using the old Python bindings of ORBit with the 0.x release of ORBit, there's no communciation happening at all. I'm going to try to write the client in C, using ORBit 2.5.0 and earlier stable releases, to try to isolate the problem. But, if it is a known problem, or if anyone has a solution, please tell me :-) With friendly regards, Takis |
From: Christian R. <ki...@as...> - 2002-11-22 19:10:58
|
On Fri, Nov 22, 2002 at 04:14:32PM +0100, Panagiotis Issaris wrote: > I'm using Debian GNU/Linux with orbit-python 1.99.0-4 on Python 2.2 with > ORBit version 2.5.0. I'm using MICO 2.3.7. The MICO C++ server works, > and also works with a MICO C++ client. The ORBit Python client works > partly. Hello Panagiotis, The author of the port you are using was Johan Dahlin, but he has unsubscribed from the mailing list and is refusing to answer emails about it to anybody. I would suggest you tried to use James Henstridge's version, called pyorbit, which is also based on GNOME2. You can get a recent version at: http://ftp.gnome.org/pub/GNOME/sources/pyorbit/1.99/pyorbit-1.99.2.tar.gz O-P is quite dead, and because other development efforts are in order, I'm not sure it will ever resurrect. Take care, -- Christian Reis, Senior Engineer, Async Open Source, Brazil. http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL |
From: Jason T. <ta...@au...> - 2002-11-22 21:38:18
|
Hi Kiko :) On Fri, 2002-11-22 at 14:10, Christian Reis wrote: > O-P is quite dead, and because other development efforts are in order, > I'm not sure it will ever resurrect. With James hacking on pyorbit, and my current "real world" life, which is insane beyond description, I have no inclination or time to do any future development on O-P. I'm a bit relieved that James is working on pyorbit. His track record is much better than mine. :) O-P is pretty stable with ORBit 0.5 AFAIK, and I'm happy to answer questions with ORBit-Python 0.3.1. So, yes, I would say that O-P has reached the end of its life cycle. Best, Jason. -- Jason Tackaberry :: ta...@au... :: 705-949-2301 x330 Academic Computing Support Specialist Information Technology Services Algoma University College :: www.auc.ca |
From: Roland M. <99....@ai...> - 2002-11-25 13:16:58
|
Jason Tackaberry (2002-11-22 16:38:16 -0500) : > On Fri, 2002-11-22 at 14:10, Christian Reis wrote: >> O-P is quite dead, and because other development efforts are in order, >> I'm not sure it will ever resurrect. [...] > So, yes, I would say that O-P has reached the end of its life cycle. Glad to see the "old-timers" on this list again... Although I am also a bit sad to hear it openly told, I agree O-P is something that has been. Since everybody seems to agree on that, I suggest that we gracefully retire the program. It might be interesting to submit O-P to the Unmaintained Free Software page[1], with contact info so that people can "adopt" it if they have some need for it. I sometimes consider unsubscribing from this mailing-list too, but since the traffic is so low and the spam seems to be mostly caught by Spamassassin, I'll probably refrain from it for now. Just in case. Roland. [1] http://www.unmaintained-free-software.org/ --=20 Roland Mas La menace de la baffe p=E8se plus lourd que la baffe elle-m=EAme. -- in Sri Raoul le petit yogi (Gaudelette) |
From: Christian R. <ki...@as...> - 2002-11-25 13:20:51
|
On Mon, Nov 25, 2002 at 02:16:52PM +0100, Roland Mas wrote: > Jason Tackaberry (2002-11-22 16:38:16 -0500) : > > > On Fri, 2002-11-22 at 14:10, Christian Reis wrote: > >> O-P is quite dead, and because other development efforts are in order, > >> I'm not sure it will ever resurrect. > > [...] > > > So, yes, I would say that O-P has reached the end of its life cycle. > > Glad to see the "old-timers" on this list again... Although I am > also a bit sad to hear it openly told, I agree O-P is something that > has been. Since everybody seems to agree on that, I suggest that we > gracefully retire the program. It might be interesting to submit O-P > to the Unmaintained Free Software page[1], with contact info so that > people can "adopt" it if they have some need for it. I guess we should ask Tack to also update o-p.sault.org and maybe freshmeat/sf.net to let people know of this beforehand, since they have a right to know the software is unmaintained. But he'll probably only read this email next april or so :-) Take care, -- Christian Reis, Senior Engineer, Async Open Source, Brazil. http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL |
From: Jason T. <ta...@au...> - 2002-11-25 14:32:35
|
On Mon, 2002-11-25 at 08:20, Christian Reis wrote: > I guess we should ask Tack to also update o-p.sault.org and maybe > freshmeat/sf.net to let people know of this beforehand, since they have > a right to know the software is unmaintained. But he'll probably only > read this email next april or so :-) Hey now. :) (Well, that was a fair jab, actually.) I'm hesitant to formally retire O-P while ORBit 0.5 is still floating around. Is ORBit 0.5.x still being maintained for bug fixes? I'd not be opposed to dealing with any bugs in ORBit-Python 0.3.1 that crop up, but I won't add new features or work on the ORBit2 port. Actually Johan had done most (all?) of the work on the ORBit2 port, and I'd probably not recognize most of the code anymore. I did have a chance to look a bit at pyorbit, though, and it does seem like James is taking it in a different direction than my vision for O-P was. It requires IDL compiling in most cases, and it seems to have a number of non-standard requirements like ORBit.load_typelib(). I went through great pains to make things as spec-compliant as possible (IDL discovery anyone?) Mind you, it was a quick look so I could be off. :) Nevertheless, working on O-P no longer scratches any itches for me. I think it probably is a good idea to update the website to indicate that there will be no new features in O-P and that it will be going away when ORBit 0.5 does. I don't want to speak about O-P 1.99 though. That's more Johan's baby, and it should be his call if he wants to put it up for adoption. Funny enough, O-P 1.99 got included in RedHat 8.0 and it is by default broken. :) Jason. -- Jason Tackaberry :: ta...@au... :: 705-949-2301 x330 Academic Computing Support Specialist Information Technology Services Algoma University College :: www.auc.ca |
From: Christian R. <ki...@as...> - 2002-11-25 14:42:07
|
On Mon, Nov 25, 2002 at 09:32:52AM -0500, Jason Tackaberry wrote: > ORBit 0.5 does. I don't want to speak about O-P 1.99 though. That's > more Johan's baby, and it should be his call if he wants to put it up > for adoption. Funny enough, O-P 1.99 got included in RedHat 8.0 and it > is by default broken. :) Johan has unsubscribed from the list and rudely refuses to answer any emails about it, so I would probably add a notice about this to wheverever you will post this update. RedHat 8.0 also included PyGTK-1.99.x (while still in beta) as default even though it broke all pre-existing pygtk programs, so judge yourself the quality of it as a distro. Take care, -- Christian Reis, Senior Engineer, Async Open Source, Brazil. http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL |
From: Jason T. <ta...@au...> - 2002-11-25 15:47:16
|
On Mon, 2002-11-25 at 09:41, Christian Reis wrote: > Johan has unsubscribed from the list and rudely refuses to answer any > emails about it, so I would probably add a notice about this to > wheverever you will post this update. Is he refusing or just ignoring? I'd give him the benefit of the doubt if you're just not hearing from him. He may not mean anything rude by it. I know I don't, and you know how bad I am for dropping off planet Earth. > RedHat 8.0 also included PyGTK-1.99.x (while still in beta) as default > even though it broke all pre-existing pygtk programs, so judge yourself > the quality of it as a distro. Well, for all its faults (and there are lots, like significant rpm(8) brokenness), I rather like RedHat 8.0. Its GNOME2 configuration is pretty nice, I think. Jason. -- Jason Tackaberry :: ta...@au... :: 705-949-2301 x330 Academic Computing Support Specialist Information Technology Services Algoma University College :: www.auc.ca |
From: Panagiotis I. <tak...@sk...> - 2002-11-25 17:44:23
|
Op vr 22-11-2002, om 20:10 schreef Christian Reis: > The author of the port you are using was Johan Dahlin, but he has > unsubscribed from the mailing list and is refusing to answer emails > about it to anybody. > > I would suggest you tried to use James Henstridge's version, called > pyorbit, which is also based on GNOME2. You can get a recent version at: > > http://ftp.gnome.org/pub/GNOME/sources/pyorbit/1.99/pyorbit-1.99.2.tar.gz > > O-P is quite dead, and because other development efforts are in order, > I'm not sure it will ever resurrect. Oh, I'm kinda disappointed to hear that. I really liked O-P's way of generating the stubs and skeletons from the IDL-file at runtime. Furthermore, I'm a bit worried about PyOrbit because the README file says: "Where it makes sense, this binding tries to stay compatible with the Python language mapping". So, from reading that I would understand that it is not? Thanks for your advice! With friendly regards, Takis -- Panagiotis Issaris <tak...@sk...> |
From: Christian R. <ki...@as...> - 2002-11-25 16:08:24
|
On Mon, Nov 25, 2002 at 10:47:33AM -0500, Jason Tackaberry wrote: > On Mon, 2002-11-25 at 09:41, Christian Reis wrote: > > Johan has unsubscribed from the list and rudely refuses to answer any > > emails about it, so I would probably add a notice about this to > > wheverever you will post this update. > > Is he refusing or just ignoring? I'd give him the benefit of the doubt > if you're just not hearing from him. He may not mean anything rude by > it. I know I don't, and you know how bad I am for dropping off planet > Earth. Refusing is the correct term. I asked him if he would answer the request sent to the lest and he said "I unsubscribed from the list for a reason"; knowing him (you know he's moved to Brazil, I guess?) quite well that means he's not interested. Take care, -- Christian Reis, Senior Engineer, Async Open Source, Brazil. http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL |
From: Jason T. <ta...@au...> - 2002-11-25 16:16:46
|
On Mon, 2002-11-25 at 11:07, Christian Reis wrote: > Refusing is the correct term. I asked him if he would answer the request > sent to the lest and he said "I unsubscribed from the list for a > reason"; knowing him (you know he's moved to Brazil, I guess?) quite > well that means he's not interested. Hrmm, doesn't sound like Johan. I wonder if he got pissed off at me for me disappearing for a while? Jason. -- Jason Tackaberry :: ta...@au... :: 705-949-2301 x330 Academic Computing Support Specialist Information Technology Services Algoma University College :: www.auc.ca |