Re: A commercial chance for using ORBit2/C++ bindings

[Bowie Owens <bowie owens csiro au>]

Yu Jie wrote:

> Hi,
>
>    Our organization is planning a big project which
> is a Linux-Embedded system and will be sold as
> products in the market.
>    It is necessary for us to use CORBA in the
> implementation for the interoperability between C++
> and Java. We are very interested in choosing
> ORBit2/C++ bindings CORBA product, so we have a few
> important questions to ask though some of the answers
> to them might be found in the website but we are a
> little afraid whether the information might be
> out-of-date. 
>    We will really appreciate your help if you could
> help us find the answers to the following questions:
>  
I'm the maintainer for the C++ bindings. So I'll try and answer your 
questions as the relate to the C++ binding. However, some of the answers 
depend on the behaviour of the C binding. This is because the C++ 
binding is implemented as wrapper around the C binding.

> 1. What's the exact latest stable version of
> ORBit2/C++ bindings?
>  
The latest stable version is 1.3.9. Releases are available at:

ftp://ftp.gnome.org/pub/gnome/sources/orbitcpp/

> 2. Which version of the CORBA specification does it
> support? (For example, CORBA 2.6)
>  
The C++ binding is a wrapper around the C binding and as such targets 
the same spec as ORBit2 which is 2.4.

> 3. Is it suitable for embedded system development?
>  
I've not done any embedded system development myself so I'm not abreast 
of the issues. The C++ binding is a wrapper as such there is a higher 
memory overhead attached since some objects are represented twice or 
must be converted. If memory consumption and execution time are really 
important, you might be better off using the C binding directly.

> 4. Has it been tested on the Linux Kernel 2.6 and
> what's the result? Namely, could it be used without
> any problem in the environment of Linux Kernel 2.6? 
>   And if it can't, what's the latest version of Linux
> Kernel does it support without any problem?
>  
The C++ binding uses the C binding for all the communication. As such it 
works where the C binding works. Someone with more familarity with the C 
binding can answer this with authority, but I don't think the C binding 
requires a specific kernel version as it runs on several different 
flavours of UNIX.

> 5. Whether there is any other software which
> ORBit2/C++ bindings must depend on to work normally?
The C++ binding depends on the C binding. Which depends on the GNOME 
libraries libIDL and glib.

>  
>
> 6. Does it well support the following features or
> CORBA services:
>   (1) SINGLE_THREAD_MODEL (POA's thread policy)
>  
The threading model is dealt with in the C binding. But if I understand 
correctly the single thread model is the default behaviour and works well.

>   (2) Valuetype
>  
No.

>   (3) Name Service
>  
Yes. The C++ binding provides a compiled version of CosNaming.idl for 
use in programs that want to connect to a name sever. The C binding 
provides an implementation of the name server I believe.

>   (4) Transport layer's replacement
>  
No.

>   (5) CORBA Component Model(CCM), Real Time,
> Event/Notification Service, Dynamic Invocation
> Interface(DII), Dynamic Skeleton Interface(DSI),
> Security Service.
>  
No. DII and DSI are available with the C binding but I don't think they 
are wrappered properly in the C++ binding.


--
Bowie Owens

CSIRO Mathematical & Information Sciences
phone  : +61 3 9545 8055
fax    : +61 3 9545 8080
mobile : 0425 729 875
email  : Bowie Owens csiro au