Why would you disable double click on the text? what too with so many complicated words in the article ?
Thu, 09 Jan 2014 06:41:28 UTC
I dropped a tear in the ocean of distributed technologies where CORBA should have been forever.
Mon, 09 Dec 2013 14:26:50 UTC
The interesting question is not about the fall of CORBA, but the rise of JEE. How could EJB 1.0 - particularly CMP entity beans - be considered a serious technology and be used in real applications? How could this push a working and well performing technology aside? it felt like an academic toy. Many years later now it is becoming usable.
Mon, 04 Nov 2013 14:25:32 UTC
some ORB implementations were developed to use CORBA not only on processors, but only on "specialized HW" like DSP, FPGA or ASIC.
Why not bringing CORBA to live to mine bitcoins ?!?
Thu, 26 Sep 2013 18:03:55 UTC
The CORBA community did agree that the C++ mapping is hard to use. In order to resolve this the OMG has standardized a new C++11 language mapping which is easy to use, reuses STL as much as possible, and leads to more performance.
Thu, 26 Sep 2013 18:02:38 UTC
The article mentions the lack of compression, this is not the case anymore. CORBA now has ZIOP which defines a set of policies to enable compression when sending data over the wire.
Wed, 21 Mar 2012 02:23:57 UTC
Anyone interested in old CORBA books let me know -- soon. Else, they are headed for the trash heap. Here's a listing:
-- Essential CORBA (Thomas Mowbray)
-- Java Prgramming with CORBA (Andreas Vogel, Keith Duddy)
-- Inside DCOM (Eddon)
-- Object Oriented Modeling and Design (James Rumbaugh, et al)
-- Component Software, Beyond Object Oriented Programming (Clemens Szyperski)
Thu, 02 Sep 2010 12:58:11 UTC
I just started using it! In 2010. Seems still very alive.
Wed, 16 Jun 2010 00:23:23 UTC
Since CORBA is pronunced in polish 'korba' which means (in polish) 'crank' it cannot be good.
Sat, 10 Apr 2010 01:43:47 UTC
AMQP seems to be showing a lot of promise as well.
Sat, 26 Dec 2009 02:12:44 UTC
CORBA has not made any mistakes. You did.
Wed, 11 Nov 2009 14:36:55 UTC
IMHO, the whole article is tendentious. The impression about CORBA being complex comes from implementation guides, not from API ones. It's just like learning basic programming from compiler design books. In most cases the application programmer interacts with CORBA like this:
CORBA::ORB_var orb = CORBA::ORB_init(argc, argv);
CORBA::Object_var objref = orb->string_to_object( "corbaloc:iiop:somehost:12121/Your/Server/Object" );
YOUR::Server_var varObject = YOUR::Server::_narrow( objref );
If one calls this complex, I really don't know what a simple thing looks like. Compare this to boiler-plated COM spagetti code, for example.
For me CORBA it the only way of making complex heterogeneous distributed systems without shooting yourself in the leg, and it has been proven to be true several times in my programming career. CORBA has it's own deficiencies, but they don't lie where the author points to. The claims of the author look at best strange given that he "worked on CORBA as a member of the Object Management Group's architecture board and as an ORB implementer, consultant, and trainer".
Thu, 01 Oct 2009 02:27:45 UTC
CORBA has fundamental problems:
- why bother hiding complexity of distributed objects if it cannot be hidden? example: if object A calls object B on a single machine, this call fails only when someone pulled a CPU from a motherboard. Bothering with handling this case is not practical. On the other hand, in a distributed system I need to be bothered ... and so much for hiding the distribution.
- asynchronous message passing is preferable to RPC for integration purposes for reasons that are too plenty to enumerate. Of course CORBA can do asynchrony, but it is an artificial afterhought.
- interfaces for business interactions are much more flexibly built using documents and not rigid object structures. I don't want to rebuild my clients each time an IDL needs to change. Also I don't want to use object / domain modeling to specify business interfaces (on that token XML Schema is bizarre, using object concepts; e.g inheritance to specify content and not behaviour!). I want loosely structured, often superfluous, specified in terms of syntax (i.e. RelaxNG for example) documents that I can access in a structure-agnostic fashion (say XPath or other XML query language) so when someone adds e-mail to customer structure or moves it to account structure, there is very little ripple effect.
So once I have to be aware of the fact that my system is distributed, I'll go message-oriented, asynchronous in most cases, REST-full perhaps, or SOAP with *no* SOAP encoding. I leave RPC to where it is appropriate (e.g technical not business interfaces) and for that I don't need the whole CORBA baggage.
Really, there is no reason for CORBA to exist. It is dead.
Fri, 25 Sep 2009 00:35:00 UTC
I'll mourne the passing of CORBA like I mourned the death of Saddam Hussein at the end of a hangman's noose.
Unfortunately, I have to connect my application to a legacy CORBA application. Someone will have to dust off some old CORBA books and figure out how this relic can be revived again. Pour me another one, sam.
Wed, 18 Feb 2009 19:49:10 UTC
Fine. CORBA goes. What next? Is someone developing real middleware that can work across different OS like windows and Linux?
Sat, 14 Feb 2009 15:02:23 UTC
In my opinion the most important reason that CORBA failed is because it did
not embrace the web and required additional ports to be opened (other than
the 80 port for HTTP which is commonly opened for HTTP conversations).
Thu, 05 Feb 2009 22:51:16 UTC
Hi Bert, sorry for the late reply--I only just now saw your post.
CORBA is built into many existing long-lived systems, so it will be around for some time yet. (Think of COBOL applications that, thirty years later, are still going.) But I believe that CORBA is essentially dead for new developments. It might make sense to use CORBA in an environment with pre-existing CORBA infrastructure but, otherwise, I don't believe there is much going on anymore. And starting a new development with CORBA now would seem a rather risky investment, seeing that there are almost no commercial vendors left, and that activity around the open source versions seems to be diminishing too.
If you want an alternative to CORBA, have a look at ZeroC's Ice. It does everything CORBA does and more, does it much more simply, runs on just about everything under the sun, and supports a large number of programming languages.
Fri, 19 Dec 2008 20:02:25 UTC
It's been quite a while since you wrote this and it's even longer since I used CORBA but I always like it and am wondering the following:
Will CORBA live on? Do you think it would be wise to use it in new designs? If not, what then?