DDS and Ada

DDS and Ada

[Per Sandberg]

We are in the process of producing a thick Ada2005 binding to DDS the 
current plan is to write a binding that resembels of the Java bainding 
specified by OMG.
And the binding shall only depend on the C libraries for DDS as 
specified by OMG.

Artifacts to be produced:
* IDL to Ada generator
* Support packages for the generated code.
* Bindings as specified by the DDS.idl file.
     The curren path is to use the idl2Ada specification for corba and
     and then tweek the bindings to resemble of the DDS Java interface.

Are ther any other parties that got intrests in this direction ???


/Per

DDS and Ada

[Martyn Pike]

Hi.  Has anyone that frequencts this group ever successfully deployed 
an Ada based DDS systems ?  Either using a COTS or custom 
implementation.

thanks.

Re: DDS and Ada

[anon]

Last DOS version of Ada is GNAT 3.10p (Ada 95) using 32-bit DJGPP 
subsystem. And since Microsoft drop support for DOS a few years back, 
its going to be hard to find someone to support Ada 2005 as well as 
Ada 2012 for DOS.

But before giving up hope you might check with DJGPP maintainers 
to see if either GNAT Ada 2005 or 2012 will be available.


There are other system, but mostly I assume Ada 2012 will be support in 
32-bit and 64-bit versions for 

  Linux
  OS/X
  Windows 7

with ports for

  Java (VM)


In <2011070520341767939-martynatemconukdotcom@news-europe.giganews.com>, Martyn Pike <martyn at emconuk dot com> writes:
>Hi.  Has anyone that frequencts this group ever successfully deployed 
>an Ada based DDS systems ?  Either using a COTS or custom 
>implementation.
>
>thanks.
>

Re: DDS and Ada

[Maciej Sobczak]

On Jul 5, 9:34 pm, Martyn Pike <martyn at emconuk dot com> wrote:

> Hi.  Has anyone that frequencts this group ever successfully deployed
> an Ada based DDS systems ?  Either using a COTS or custom
> implementation.

Do you insist on the exact DDS implementation or is your interest of
wider scope?
If you really need DDS, then I think RTI has an Ada binding to one of
their products.
Otherwise you might also want to have a look at this:

http://www.inspirel.com/yami4/

This system can be used as a lightweight alternative to DDS,
especially with its flexible message broker that supports publish-
subscribe with hierarchic multiway message matching. As a side note,
this broker was actually implemented in Ada, which is unlike most
(all?) messaging products out there that merely offer a thin binding
or a client interface to something implemented in C or Java.

In either case, don't hesitate to post your findings or conclusions
here, the subject of distributed systems is very much Ada-related.

--
Maciej Sobczak * http://www.msobczak.com * http://www.inspirel.com

Re: DDS and Ada

[anon]

Appendment try 

  http://ap1.pp.fi/djgpp/gcc/

In <2011070520341767939-martynatemconukdotcom@news-europe.giganews.com>, Martyn Pike <martyn at emconuk dot com> writes:
>Hi.  Has anyone that frequencts this group ever successfully deployed 
>an Ada based DDS systems ?  Either using a COTS or custom 
>implementation.
>
>thanks.
>

Re: DDS and Ada

[Frank J. Lhota]

On 7/5/2011 3:34 PM, Martyn Pike wrote:
> Hi. Has anyone that frequencts this group ever successfully deployed an
> Ada based DDS systems ? Either using a COTS or custom implementation.
>
> thanks.
>

The acronym DDS is heavily overloaded. What does DDS stand for? (BTW I'm 
fairly sure that you did not mean DOS).

-- 
"All things extant in this world,
Gods of Heaven, gods of Earth,
Let everything be as it should be;
Thus shall it be!"
- Magical chant from "Magical Shopping Arcade Abenobashi"

"Drizzle, Drazzle, Drozzle, Drome,
Time for this one to come home!"
- Mr. Wizard from "Tooter Turtle"

Re: DDS and Ada

[Martyn Pike]

On 2011-07-05 22:58:57 +0100, Frank J. Lhota said:

> The acronym DDS is heavily overloaded. What does DDS stand for? (BTW 
> I'm fairly sure that you did not mean DOS).

Yes - I meant the OMG Data Distribution Service and not DOS.

Re: DDS and Ada

[anon]

You only have Polyorb for GNAT

     polyorb-gpl-2011-src.tgz

The problem is that polyorb is a third party package that is not governed by 
the RM.

Sorry, about DOS but after a numbers of hours days viewing monitor screen
sometimes a "DDS" can look like "DOS".

In <2011070619460659694-martynatemconukdotcom@news-europe.giganews.com>, Martyn Pike <martyn at emconuk dot com> writes:
>On 2011-07-05 22:58:57 +0100, Frank J. Lhota said:
>
>> The acronym DDS is heavily overloaded. What does DDS stand for? (BTW 
>> I'm fairly sure that you did not mean DOS).
>
>Yes - I meant the OMG Data Distribution Service and not DOS.
>

Re: DDS and Ada

[Martyn Pike]

On 2011-07-05 21:59:54 +0100, Maciej Sobczak said:

> On Jul 5, 9:34�pm, Martyn Pike <martyn at emconuk dot com> wrote:
> 
>> Hi. �Has anyone that frequencts this group ever successfully deployed
>> an Ada based DDS systems ? �Either using a COTS or custom
>> implementation.
> 
> Do you insist on the exact DDS implementation or is your interest of
> wider scope?
> If you really need DDS, then I think RTI has an Ada binding to one of
> their products.
> Otherwise you might also want to have a look at this:
> 
> http://www.inspirel.com/yami4/
> 
> This system can be used as a lightweight alternative to DDS,
> especially with its flexible message broker that supports publish-
> subscribe with hierarchic multiway message matching. As a side note,
> this broker was actually implemented in Ada, which is unlike most
> (all?) messaging products out there that merely offer a thin binding
> or a client interface to something implemented in C or Java.
> 
> In either case, don't hesitate to post your findings or conclusions
> here, the subject of distributed systems is very much Ada-related.

My interest is in the wider scope but specifically the interoperability 
of different solutions.
I understand that different CORBA ORBs can talk to eachother but I am 
less convinced
different DDS implementations can talk to eachother.  So a C++ DDS 
publisher and an
Ada DDS subscriber as per the spec is what I am looking for.  Ada DDS 
implementations seem
thin on the ground

Re: DDS and Ada

[Martyn Pike]

On 2011-07-06 20:12:09 +0100, anon@att.net said:

> You only have Polyorb for GNAT
> 
>      polyorb-gpl-2011-src.tgz
> 
> The problem is that polyorb is a third party package that is not governed by
> the RM.
> 
> Sorry, about DOS but after a numbers of hours days viewing monitor screen
> sometimes a "DDS" can look like "DOS".
> 
> In 
> <2011070619460659694-martynatemconukdotcom@news-europe.giganews.com>, 
> Martyn Pike <martyn at emconuk dot com> writes:
>> On 2011-07-05 22:58:57 +0100, Frank J. Lhota said:
>> 
>>> The acronym DDS is heavily overloaded. What does DDS stand for? (BTW
>>> I'm fairly sure that you did not mean DOS).
>> 
>> Yes - I meant the OMG Data Distribution Service and not DOS.

and PolyORB only seems to do the CORBA from OMG and not DDS.

No problem about the DOS thing.

Re: DDS and Ada

[Dmitry A. Kazakov]

On Wed, 6 Jul 2011 19:46:06 +0100, Martyn Pike wrote:

> Yes - I meant the OMG Data Distribution Service and not DOS.

We have implemented such a system in Ada 2005 (GNAT Pro). It is presently 
in the testing phase.

The system is more than just DDS, it is a middleware where distribution is 
one feature among others. Another is device/protocols abstraction.

For distribution the middleware supports our proprietary TCP/IP-based 
protocol. Possibly we also will support our PGM-based (multicast) protocol 
if there will be demand. For the end devices we support the XCP protocol 
(both masters and slaves) and the EtherCAT slaves (it is almost ready, now 
is under test).

The middleware is designed for tight real-time constraints. In particular, 
our customers demand 10kHz sampling and distribution without data loss. 
Last week we demonstrated such a test setup.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: DDS and Ada

[Maciej Sobczak]

On Jul 6, 9:27 pm, Martyn Pike <martyn at emconuk dot com> wrote:

> My interest is in the wider scope but specifically the interoperability
> of different solutions.
> I understand that different CORBA ORBs can talk to eachother but I am
> less convinced
> different DDS implementations can talk to eachother.

Not necessarily. If they can, their vendors make a big marketing
hoopla about it, which means that it is not expected out of the box.
In particular, I do not expect broker-based DDS to interoperate with a
broker-less DDS without any additional gateway/bridge on the way.

> So a C++ DDS
> publisher and an
> Ada DDS subscriber as per the spec is what I am looking for.

Then probably the best choice is to find a vendor who has a single
product with several language bindings.

May I ask you why do you insist on DDS, specifically? Is it mandated
by some higher authority or is it some very special feature/
requirement that can only be possibly met by a DDS implementation?

--
Maciej Sobczak * http://www.msobczak.com * http://www.inspirel.com

Re: DDS and Ada

[Shark8]

On Jul 6, 2:12 pm, a...@att.net wrote:
> You only have Polyorb for GNAT
>
>      polyorb-gpl-2011-src.tgz
>
> The problem is that polyorb is a third party package that is not governed by
> the RM.
>
> Sorry, about DOS but after a numbers of hours days viewing monitor screen
> sometimes a "DDS" can look like "DOS".
>
> In <2011070619460659694-martynatemconukdot...@news-europe.giganews.com>, Martyn Pike <martyn at emconuk dot com> writes:
>
>
>
>
>
>
>
> >On 2011-07-05 22:58:57 +0100, Frank J. Lhota said:
>
> >> The acronym DDS is heavily overloaded. What does DDS stand for? (BTW
> >> I'm fairly sure that you did not mean DOS).
>
> >Yes - I meant the OMG Data Distribution Service and not DOS.

Though a DOS Ada 2012 compiler might be very nice for one wishing to
do an OS.

Re: DDS and Ada

[anon]

If you know what your doing you can use Linux, or Apple OS/X (unsure 
about Windows) version of GNAT and create a nice OS (32 or 64 bit).   

Even a small Linux Kernel, written in Ada is possible. Of course, the 
full Linux 3.x kernel might take some time.

With DJGPP GCC 4.6x, it might work.

In <5b042e9c-7db8-4a20-aa87-8daa764e50f9@x16g2000yqg.googlegroups.com>, Shark8 <onewingedshark@gmail.com> writes:
>On Jul 6, 2:12=A0pm, a...@att.net wrote:
>> You only have Polyorb for GNAT
>>
>> =A0 =A0 =A0polyorb-gpl-2011-src.tgz
>>
>> The problem is that polyorb is a third party package that is not governed=
> by
>> the RM.
>>
>> Sorry, about DOS but after a numbers of hours days viewing monitor screen
>> sometimes a "DDS" can look like "DOS".
>>
>> In <2011070619460659694-martynatemconukdot...@news-europe.giganews.com>, =
>Martyn Pike <martyn at emconuk dot com> writes:
>>
>>
>>
>>
>>
>>
>>
>> >On 2011-07-05 22:58:57 +0100, Frank J. Lhota said:
>>
>> >> The acronym DDS is heavily overloaded. What does DDS stand for? (BTW
>> >> I'm fairly sure that you did not mean DOS).
>>
>> >Yes - I meant the OMG Data Distribution Service and not DOS.
>
>Though a DOS Ada 2012 compiler might be very nice for one wishing to
>do an OS.

Re: DDS and Ada

[Martyn Pike]

On 2011-07-06 22:24:21 +0100, Maciej Sobczak said:
> 
> May I ask you why do you insist on DDS, specifically? Is it mandated
> by some higher authority or is it some very special feature/
> requirement that can only be possibly met by a DDS implementation?

I am interested in doing the OMG Certificationg for Realtime and 
Embedded Systems.  They focus specifically on DDS.  I have no specific 
requirement but I am always on the lookout for reusable solutions as 
much of the work I do is in mission critical distributed systems.

I will definately look at yami4 - and may even have a read of your 
book. looks very good.

Re: DDS and Ada

[Gautier write-only]

On 7 juil, 04:34, Shark8 <onewingedsh...@gmail.com> wrote:

> Though a DOS Ada 2012 compiler might be very nice for one wishing to
> do an OS.

Really?
Anyway, if Ada 2012 is supported by GCC 4.6.1, you have your DOS Ada
2012 compiler:

http://groups.google.com/group/comp.os.msdos.djgpp/browse_frm/thread/8d09df68213d446a#

It seems to work pretty well:
http://oi56.tinypic.com/6sz87q.jpg
______________________________________________________________________________
Gautier's Ada programming -- http://gautiersblog.blogspot.com/search/label/Ada
NB: follow the above link for a valid e-mail address

Re: DDS and Ada

[Shark8]

On Jul 6, 10:23 pm, a...@att.net wrote:
> If you know what your doing you can use Linux, or Apple OS/X (unsure
> about Windows) version of GNAT and create a nice OS (32 or 64 bit).  
>
> Even a small Linux Kernel, written in Ada is possible. Of course, the
> full Linux 3.x kernel might take some time.

Er... Why would one want to write Linux in Ada?
Linux seems to me to have a philosophy directly opposite to Ada's.

> With DJGPP GCC 4.6x, it might work.

I guess I'll have to take a look at it.

Re: DDS and Ada

[Shark8]

On Jul 7, 8:04 pm, Gautier write-only <gautier_niou...@hotmail.com>
wrote:
> On 7 juil, 04:34, Shark8 <onewingedsh...@gmail.com> wrote:
>
> > Though a DOS Ada 2012 compiler might be very nice for one wishing to
> > do an OS.
>
> Really?

DOS is so minimal/unconstrained that one can easily make bootable
program with a DOS compiler. I had the start of an OS (I got it to the
point of being able to recognize commands and change video-modes)
written purely in BP7 {100% Pascal, with the exception of something
like 6-lines of inline assembler (to handle the keyboard/scan-codes)}.

The problem currently stumping me regarding doing it in Ada is that my
GNAT tries to hook the non-existent runtime OR (with the runtime-
restriction pragma) gives me funny linker errors preventing a compile.


> Anyway, if Ada 2012 is supported by GCC 4.6.1, you have your DOS Ada
> 2012 compiler:
>
> http://groups.google.com/group/comp.os.msdos.djgpp/browse_frm/thread/...

Oh, thanks for the lead.

Re: DDS and Ada

[Martyn Pike]

On 2011-07-06 21:13:53 +0100, Dmitry A. Kazakov said:

> On Wed, 6 Jul 2011 19:46:06 +0100, Martyn Pike wrote:
> 
>> Yes - I meant the OMG Data Distribution Service and not DOS.
> 
> We have implemented such a system in Ada 2005 (GNAT Pro). It is presently
> in the testing phase.
> 
> The system is more than just DDS, it is a middleware where distribution is
> one feature among others. Another is device/protocols abstraction.
> 
> For distribution the middleware supports our proprietary TCP/IP-based
> protocol. Possibly we also will support our PGM-based (multicast) protocol
> if there will be demand. For the end devices we support the XCP protocol
> (both masters and slaves) and the EtherCAT slaves (it is almost ready, now
> is under test).
> 
> The middleware is designed for tight real-time constraints. In particular,
> our customers demand 10kHz sampling and distribution without data loss.
> Last week we demonstrated such a test setup.

This sounds very interesting indeed.  Will any of this be published 
publically ?
I appreciated it's your companies valuable IP but next years Ada Europe 
conference would benefit from a presentation on this.

Martyn.

Re: DDS and Ada

[Gautier write-only]

For Shark8: DJGPP (then GNAT/GCC 4.6.1) and Ada programs compiled with
DJGPP runs on a DPMI-extended DOS.
It can be
* MS-DOS, FreeDOS, NTDVM, DR-DOS without multitasking plus CWSDPMI
* DR-DOS with multitasking, which provide its own DPMI

If you want an Ada/DJGPP example with keyboard, sound, and video
drivers written in Ada, you can check this:
http://sites.google.com/site/rugxulo/eng3d018.zip

G.

Re: DDS and Ada

[Dmitry A. Kazakov]

On Fri, 8 Jul 2011 06:33:37 +0100, Martyn Pike wrote:

> On 2011-07-06 21:13:53 +0100, Dmitry A. Kazakov said:
> 
>> On Wed, 6 Jul 2011 19:46:06 +0100, Martyn Pike wrote:
>> 
>>> Yes - I meant the OMG Data Distribution Service and not DOS.
>> 
>> We have implemented such a system in Ada 2005 (GNAT Pro). It is presently
>> in the testing phase.
>> 
>> The system is more than just DDS, it is a middleware where distribution is
>> one feature among others. Another is device/protocols abstraction.
>> 
>> For distribution the middleware supports our proprietary TCP/IP-based
>> protocol. Possibly we also will support our PGM-based (multicast) protocol
>> if there will be demand. For the end devices we support the XCP protocol
>> (both masters and slaves) and the EtherCAT slaves (it is almost ready, now
>> is under test).
>> 
>> The middleware is designed for tight real-time constraints. In particular,
>> our customers demand 10kHz sampling and distribution without data loss.
>> Last week we demonstrated such a test setup.
> 
> This sounds very interesting indeed.  Will any of this be published 
> publically ?

It is planned, but I don't know when.

> I appreciated it's your companies valuable IP but next years Ada Europe 
> conference would benefit from a presentation on this.

It is difficult to find time for conferences. Maybe, but I cannot say
anything for sure.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: DDS and Ada

[Randy Brukardt]

"Shark8" <onewingedshark@gmail.com> wrote in message 
news:ccd482d7-1693-4540-8e88-d552caa1c126@ct4g2000vbb.googlegroups.com...
>On Jul 7, 8:04 pm, Gautier write-only <gautier_niou...@hotmail.com>
>wrote:
>> On 7 juil, 04:34, Shark8 <onewingedsh...@gmail.com> wrote:
>>
>> > Though a DOS Ada 2012 compiler might be very nice for one wishing to
>> > do an OS.
>>
>> Really?
>
>DOS is so minimal/unconstrained that one can easily make bootable
>program with a DOS compiler.

Right. Our embedded packages and DOS compilers were essentially the same 
back in the day; the only thing that DOS really did for us was load the 
program into memory (and provide a file system, but of course I/O is easily 
separable). Trap handling was done directly in the hardware, tasking was 
handled solely by our RTS, and most of the memory management (before 
Extended Memory, anyway) was done by our RTS. So all that needed to be 
different was the loader code (burn to EEPROM rather than load from disk, 
and set up a few additional registers).

                                Randy.

Re: DDS and Ada

[Randy Brukardt]

"Gautier write-only" <gautier_niouzes@hotmail.com> wrote in message 
news:71d885d1-cf1b-4d45-8b67-6a1217334f44@bl1g2000vbb.googlegroups.com...
> For Shark8: DJGPP (then GNAT/GCC 4.6.1) and Ada programs compiled with
> DJGPP runs on a DPMI-extended DOS.

I don't think that is the sort of MS-DOS that Shark8 was referring to. DPMI 
is precisely the kind of support stuff that you don't want on your bare 
machine (at least not if you are planning to write "all Ada").

You could in theory boot a 32-bit DOS executable with solely a small chunk 
of assembler code to set up protected mode, but that's hard to do (I could 
never get it to work, in large part because there is no way to find out why 
it didn't work).

                                      Randy.

Re: DDS and Ada

[Per Sandberg]

Well
I did write an Ada binding to the RTIDDS "http://www.rti.com" 
DDS-Implementation  and it used in a mixed language system,  Ada/C++/Java.
This binding is avalible as an addon to their proriarty product.
But given a resonable clean C Interface and a well structured 
codegenerator the amount of work to produce a similar binding to some 
other implementation is definitly less then 1000 hours.
The workflow I used was:

* Generate code stubs from the DDS-idl using iac from plyorb.
* Tweak the interfaces towards Ada 2005
* Generate Ada interfaces to the C-implementation using teh 
"-fdump-ada-spec in GNAT.
* Bind the things together.
* Produce a new code generator that is capable of generating Ada.
* Test
* Test
* Done
* Maintain
* ....

/Per

On 07/05/2011 09:34 PM, Martyn Pike wrote:
> Hi. Has anyone that frequencts this group ever successfully deployed an
> Ada based DDS systems ? Either using a COTS or custom implementation.
>
> thanks.
>

Re: DDS and Ada

[Oliver Kellogg]

On Friday, June 22, 2007 2:54:33 PM UTC+2, Per Sandberg wrote:
> We are in the process of producing a thick Ada2005 binding to DDS the 
> current plan is to write a binding that resembles of the Java binding 
> specified by OMG.
> And the binding shall only depend on the C libraries for DDS as 
> specified by OMG.
> 
> Artifacts to be produced:
> * IDL to Ada generator
> * Support packages for the generated code.
> * Bindings as specified by the DDS.idl file.
>      The current path is to use the idl2Ada specification for corba and
>      and then tweak the bindings to resemble of the DDS Java interface.
> 
> Are there any other parties that got interests in this direction ???
> 
> 
> /Per

Hello Per,

I realize this is a very late reply to your posting:

Is it correct that RTI have integrated your binding effort into their Ada offering?
Does your IDL to Ada generator map IDL union to Ada variant record?

Thanks,

Oliver