PolyORB 1.1r

PolyORB 1.1r

[Laurent Pautet]

Announcing the release of PolyORB 1.1r
======================================

The PolyORB team is proud to announce the release of PolyORB 1.1r 
    (http://libre.act-europe.fr/polyorb/).

This release contains a CORBA-compliant instantiation of the PolyORB
generic middleware. This release is not supported by ACT. Release 1.1r
is a stable snapshot of the PolyORB generic middleware implementing
the new features described below.

In addition to PolyORB 1.0p (2003-06-16), it includes:

* a significant increase in performance: from 30% up to 40% depending
  on the configuration,

* fixes for several bugs and memory leaks,

* extended support for CORBA and GIOP specifications,

* the PolyORB User's Guide,

* the MIOP/UIPMC protocol stack, Unreliable Multicast Inter-ORB
  Protocol, following the OMG standard,

* the DIOP protocol stack, Datagram-based Inter-ORB Protocol, a
  specialization of GIOP for oneway requests,

Other instantiations of PolyORB are available in the public PolyORB
CVS repository for testing purposes. Available instantiations include
DSA (Distributed System Annex), MOMA (Message Oriented Middleware for
Ada) and AWS (Ada Web Server).

PolyORB is primarily developed by J�r�me Hugues, Thomas Vergnaud, and
Laurent Pautet (T�l�com Paris), and Thomas Quinot (ACT Europe).
Fabrice Kordon (LIP6) also participates in the project.
Vadim Godunko regularly contributes by submitting patches.

Re: PolyORB 1.1r

[Ludovic Brenta]

Laurent Pautet writes:
> Announcing the release of PolyORB 1.1r
[...]

This is one of the things I would like to package for Debian, but I'm
not sure I have enough time to do it justice; nor am I sure of demand
for it in the Debian user community.  Any takers?  Opinions?

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Bruce Conroy]

Ludovic Brenta wrote:

> Laurent Pautet writes:
>> Announcing the release of PolyORB 1.1r
> [...]
> 
> This is one of the things I would like to package for Debian, but I'm
> not sure I have enough time to do it justice; nor am I sure of demand
> for it in the Debian user community.  Any takers?  Opinions?
> 

I'd be interested in it for Debian, but I've been trying to get the current
version to compile, and have been told that I need gcc 3.4 to do so. I
don't see that version of gcc on Debian yet.

bruce conroy

Re: PolyORB 1.1r

[Ludovic Brenta]

Bruce Conroy writes:
> Ludovic Brenta wrote:
>
>> Laurent Pautet writes:
>>> Announcing the release of PolyORB 1.1r
>> [...]
>> 
>> This is one of the things I would like to package for Debian, but I'm
>> not sure I have enough time to do it justice; nor am I sure of demand
>> for it in the Debian user community.  Any takers?  Opinions?
>> 
>
> I'd be interested in it for Debian, but I've been trying to get the current
> version to compile, and have been told that I need gcc 3.4 to do so. I
> don't see that version of gcc on Debian yet.
>
> bruce conroy

Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
it's just because of pragma Unreferenced, it is trivial to patch it to
compile with 3.15p.

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Martin Krischik]

Ludovic Brenta wrote:

> Bruce Conroy writes:
>> Ludovic Brenta wrote:
>>
>>> Laurent Pautet writes:
>>>> Announcing the release of PolyORB 1.1r
>>> [...]
>>> 
>>> This is one of the things I would like to package for Debian, but I'm
>>> not sure I have enough time to do it justice; nor am I sure of demand
>>> for it in the Debian user community.  Any takers?  Opinions?
>>> 
>>
>> I'd be interested in it for Debian, but I've been trying to get the
>> current version to compile, and have been told that I need gcc 3.4 to do
>> so. I don't see that version of gcc on Debian yet.
>>
>> bruce conroy
> 
> Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
> it's just because of pragma Unreferenced, it is trivial to patch it to
> compile with 3.15p.

The problem I found with Florist, Glade and ASIS is that most ACT Tools
compile with -gnatg which implies -gnatwa, -gnatwe and-gnaty.

A fix is easy but time consuming.

With Regards

Martin

-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: PolyORB 1.1r

[Laurent Pautet]

Martin Krischik <krischik@users.sourceforge.net> writes:

> Ludovic Brenta wrote:
>
> The problem I found with Florist, Glade and ASIS is that most ACT Tools
> compile with -gnatg which implies -gnatwa, -gnatwe and-gnaty.

This is not so time consuming. Edit configure.in. Find these
flags. Remove them, regenerate configure (with autoconf/automake). And
that's it.

It will take you less than one minute.

--
--  Laurent

Re: PolyORB 1.1r

[Bruce Conroy]

Laurent Pautet wrote:

> Martin Krischik <krischik@users.sourceforge.net> writes:
> 
>> Ludovic Brenta wrote:
>>
>> The problem I found with Florist, Glade and ASIS is that most ACT Tools
>> compile with -gnatg which implies -gnatwa, -gnatwe and-gnaty.
> 
> This is not so time consuming. Edit configure.in. Find these
> flags. Remove them, regenerate configure (with autoconf/automake). And
> that's it.
> 
> It will take you less than one minute.
> 
> --
> --  Laurent

A little background. I have been an Ada partisan since 1985 or so. I moved
many thousand lines of code from Ada 83 to Ada 95 with minor problems. I
have used the same code on Windows, Solaris and Linux. I have also moved
many lines of code from one C or C++ compiler to another, and cursed
Kernigham, Ritchie and Soustrup every time. Ada was portable.

I am starting a new project where CORBA seemed like the solution. I
downloaded AdaBroker and compiled it with GNAT 3.14p (Debian Sarge). I
needed another system to test the networking, and my test system is Fedora
Core 2, gcc 3.3.3-7 (the cutting edge as of May 18.) AdaBroker failed. I
sent a complaint, and was told "AdaBroker is obsolete, use PolyOrb." I
tried and found it won't compile under either gnat 3.14p or gcc 3.3.3-7. I
don't want to spend my time debugging tools needed to get my work done.
What has happened to the advantage Ada had over C/C++??

blc

Re: PolyORB 1.1r

[Jeff C,]

"Bruce Conroy" <bruceconroy@earthlink.net> wrote in message
news:nRPxc.8431$uX2.4847@newsread2.news.pas.earthlink.net...
>>
> A little background. I have been an Ada partisan since 1985 or so. I moved
> many thousand lines of code from Ada 83 to Ada 95 with minor problems. I
> have used the same code on Windows, Solaris and Linux. I have also moved
> many lines of code from one C or C++ compiler to another, and cursed
> Kernigham, Ritchie and Soustrup every time. Ada was portable.
>
> I am starting a new project where CORBA seemed like the solution. I
> downloaded AdaBroker and compiled it with GNAT 3.14p (Debian Sarge). I
> needed another system to test the networking, and my test system is Fedora
> Core 2, gcc 3.3.3-7 (the cutting edge as of May 18.) AdaBroker failed. I
> sent a complaint, and was told "AdaBroker is obsolete, use PolyOrb." I
> tried and found it won't compile under either gnat 3.14p or gcc 3.3.3-7. I
> don't want to spend my time debugging tools needed to get my work done.
> What has happened to the advantage Ada had over C/C++??
>
> blc
>


My guess....You went from using supported compilers and tools to bleeding
edge broken tools (gcc 3.3.3) and/or out of date tools (gnat 3.14p) with a
brand new out of the box public library.

But your point is well taken, clearly two bad experiences with hobbiest
supported tools and libraries after moving many thousands of lines indicates
the language is beyond saving.

Re: PolyORB 1.1r

[Ludovic Brenta]

Bruce Conroy wrote:
> I am starting a new project where CORBA seemed like the solution. I
> downloaded AdaBroker and compiled it with GNAT 3.14p (Debian
> Sarge). I needed another system to test the networking, and my test
> system is Fedora Core 2, gcc 3.3.3-7 (the cutting edge as of May
> 18.) AdaBroker failed. I sent a complaint, and was told "AdaBroker
> is obsolete, use PolyOrb." I tried and found it won't compile under
> either gnat 3.14p or gcc 3.3.3-7. I don't want to spend my time
> debugging tools needed to get my work done.  What has happened to
> the advantage Ada had over C/C++??

I maintain GNAT in Debian.  Debian Sarge has had GNAT 3.15p since
January 2004, and, as I noted elsewhere in this thread, you should not
have any trouble compiling PolyORB with it.

You should not use GCC 3.3.3 for Ada (my words are stronger than Ada
Core's, see [1]).  On Fedora Core, you should choose between GNAT
3.15p and GCC 3.4.  If you do not want to spent time building the
compiler, you can use Ada Core's binary distribution of 3.15p.

[1] http://libre.act-europe.fr/gnat

If you decide you want to package PolyORB for Debian, be sure to use
GNAT 3.15p, even if this requires patching PolyORB (e.g. for pragma
Unreferenced).  Reasons for this are detailed here:

http://users.skynet.be/ludovic.brenta/debian-ada-policy.html

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Martin Krischik]

Ludovic Brenta wrote:

> You should not use GCC 3.3.3 for Ada (my words are stronger than Ada
> Core's, see [1]).

Except for compiling your own gcc 3.4.1 ;-).

With Regards

Martin
-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: PolyORB 1.1r

[Bruce Conroy]

Ludovic Brenta wrote:

> Bruce Conroy wrote:
>> I am starting a new project where CORBA seemed like the solution. I
>> downloaded AdaBroker and compiled it with GNAT 3.14p (Debian
>> Sarge). I needed another system to test the networking, and my test
>> system is Fedora Core 2, gcc 3.3.3-7 (the cutting edge as of May
>> 18.) AdaBroker failed. I sent a complaint, and was told "AdaBroker
>> is obsolete, use PolyOrb." I tried and found it won't compile under
>> either gnat 3.14p or gcc 3.3.3-7. I don't want to spend my time
>> debugging tools needed to get my work done.  What has happened to
>> the advantage Ada had over C/C++??
> 
> I maintain GNAT in Debian.  Debian Sarge has had GNAT 3.15p since
> January 2004, and, as I noted elsewhere in this thread, you should not
> have any trouble compiling PolyORB with it.
> 
> You should not use GCC 3.3.3 for Ada (my words are stronger than Ada
> Core's, see [1]).  On Fedora Core, you should choose between GNAT
> 3.15p and GCC 3.4.  If you do not want to spent time building the
> compiler, you can use Ada Core's binary distribution of 3.15p.
> 
> [1] http://libre.act-europe.fr/gnat
> 
> If you decide you want to package PolyORB for Debian, be sure to use
> GNAT 3.15p, even if this requires patching PolyORB (e.g. for pragma
> Unreferenced).  Reasons for this are detailed here:
> 
> http://users.skynet.be/ludovic.brenta/debian-ada-policy.html
> 
Thanks. I installed gnat 3.15p on both systems. I now get a clean compile on
Fedora, but on Sarge I get:

../support/adacompiler -c -g -O2 -gnatfy -gnatwae -gnatpn -gnatg -I. -I.
polyorb-setup-tasking-ravenscar.ads -o polyorb-setup-tasking-ravenscar.o
+===========================GNAT BUG DETECTED==============================+
| 3.15p  (20020523) (i486-pc-linux-gnu) Assert_Failure expander.adb:83     |
| Error detected at polyorb-utils-chained_lists.adb:44:4
[polyorb-annotations.ads:79:4]|
| Please submit bug report by email to report@gnat.com.                    |
| Use a subject line meaningful to you and us to track the bug.            |
| (include your customer number #nnn in the subject line).                 |
| Include the entire contents of this bug box in the report.               |
| Include the exact gnatgcc or gnatmake command that you entered.          |
| Also include sources listed below in gnatchop format                     |
| (concatenated together with no headers between files).                   |
| (use plain ASCII or MIME attachment, or FTP to your customer directory). |
| See README.GNATPRO for full info on procedure for submitting bugs.       |
+==========================================================================+

Please include these source files with error report

polyorb-setup-tasking-ravenscar.ads
polyorb-setup-tasking.ads
polyorb-setup.ads
polyorb.ads
polyorb-orb.ads
polyorb-asynch_ev.ads
polyorb-annotations.ads
polyorb-utils.ads
polyorb-utils-chained_lists.ads
polyorb-jobs.ads
polyorb-binding_data.ads
polyorb-components.ads
polyorb-sequences.ads
polyorb-sequences-unbounded.ads
polyorb-exceptions.ads
polyorb-any.ads
polyorb-smart_pointers.ads
polyorb-tasking.ads
polyorb-tasking-mutexes.ads
polyorb-types.ads
polyorb-objects.ads
polyorb-transport.ads
polyorb-buffers.ads
polyorb-opaque.ads
polyorb-opaque-chunk_pools.ads
polyorb-sockets.ads
polyorb-sockets_initialization.ads
polyorb-filters.ads
polyorb-obj_adapters.ads
polyorb-any-nvlist.ads
polyorb-servants.ads
polyorb-poa_policies.ads
polyorb-poa_policies-thread_policy.ads
polyorb-references.ads
polyorb-utils-strings.ads
polyorb-requests.ads
polyorb-any-exceptionlist.ads
polyorb-task_info.ads
polyorb-tasking-condition_variables.ads
polyorb-tasking-threads.ads
polyorb-utils-simple_flags.ads
polyorb-scheduler.ads
polyorb-tasking-profiles.ads
polyorb-tasking-profiles-ravenscar.ads
polyorb-tasking-profiles-ravenscar-threads.ads
polyorb-tasking-profiles-ravenscar-index_manager.ads
polyorb-tasking-profiles-ravenscar-mutexes.ads
polyorb-tasking-profiles-ravenscar-condition_variables.ads
polyorb-utils-rt_calendar.ads
polyorb-calendar.ads

list may be incomplete
compilation abandoned
mv: cannot stat `tmp2078/polyorb-setup-tasking-ravenscar.o': No such file or
directory
mv: cannot stat `tmp2078/*': No such file or directory
make[2]: *** [polyorb-setup-tasking-ravenscar.lo] Error 1
make[2]: Leaving directory `/home/blc/codes/polyorb-1.1r/src'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/blc/codes/polyorb-1.1r/src'
make: *** [all-recursive] Error 1

Re: PolyORB 1.1r

[Jerome Hugues]

In article <kY%xc.489$Wr.76@newsread1.news.pas.earthlink.net>, 
Bruce Conroy wrote:

> Thanks. I installed gnat 3.15p on both systems. I now get a clean compile on
> Fedora, but on Sarge I get:

strange, both compilers should return the same result, this is odd ..

> ../support/adacompiler -c -g -O2 -gnatfy -gnatwae -gnatpn -gnatg -I. -I.
> polyorb-setup-tasking-ravenscar.ads -o polyorb-setup-tasking-ravenscar.o
> +===========================GNAT BUG DETECTED==============================+
>| 3.15p  (20020523) (i486-pc-linux-gnu) Assert_Failure expander.adb:83     |
>| Error detected at polyorb-utils-chained_lists.adb:44:4
> [polyorb-annotations.ads:79:4]|
>| Please submit bug report by email to report@gnat.com.                    |
>| Use a subject line meaningful to you and us to track the bug.            |
>| (include your customer number #nnn in the subject line).                 |
>| Include the entire contents of this bug box in the report.               |
>| Include the exact gnatgcc or gnatmake command that you entered.          |
>| Also include sources listed below in gnatchop format                     |
>| (concatenated together with no headers between files).                   |
>| (use plain ASCII or MIME attachment, or FTP to your customer directory). |
>| See README.GNATPRO for full info on procedure for submitting bugs.       |
> +==========================================================================+

PolyORB.*.Ravenscar.* packages provide tasking abstractions compatible
with the Ravenscar profile. You may consider not compiling them, as
this is of little interest for most applications.

edit src/Makefile.am 
run support/reconfig
./configure <your_options>
gmake

PS: there is a mailing list dedicated to discussion around PolyORB, see
libre.act-europe.f/polyorb for more details.

-- 
Jerome

Re: PolyORB 1.1r

[Ludovic Brenta]

Jerome Hugues wrote:
> Bruce Conroy wrote:
> 
> > Thanks. I installed gnat 3.15p on both systems. I now get a clean compile on
> > Fedora, but on Sarge I get:
> 
> strange, both compilers should return the same result, this is odd ..

Yes, odd.  Bruce, could you please file a bug in the Debian bug tracking
system?  Use reportbug, or M-x debian-bug from within emacs.  Then I'll
try to reproduce it and see if it comes from one of my patches.

> > ../support/adacompiler -c -g -O2 -gnatfy -gnatwae -gnatpn -gnatg -I. -I.
> > polyorb-setup-tasking-ravenscar.ads -o polyorb-setup-tasking-ravenscar.o

I would need to see the exact contents of ../support/adacompiler.  Is this
a shell script?

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Ludovic Brenta]

Martin Krischik wrote:
> Ludovic Brenta wrote:
> > Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
> > it's just because of pragma Unreferenced, it is trivial to patch it to
> > compile with 3.15p.
> 
> The problem I found with Florist, Glade and ASIS is that most ACT Tools
> compile with -gnatg which implies -gnatwa, -gnatwe and-gnaty.
> 
> A fix is easy but time consuming.

I do not understand what the problem is.  All three packages build
just fine on Debian with 3.15p.  By checking the buildd logs, I
have gathered the flags used:

ASIS static lib:  -g -O2 -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
ASIS shared lib:  -O2 -fPIC -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
ASIS tools:       -gnaty -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
Florist static:   -g -O2 -gnatpg
Florist shared:   -O2 -fPIC -gnatpg
Glade:            -gnatpg -O2 -gnatpn -gnatwu -gnatwe

Are you trying to say that GCC 3.4 does not support some of these flags?

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Martin Krischik]

Ludovic Brenta wrote:

> Martin Krischik wrote:
>> Ludovic Brenta wrote:
>> > Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
>> > it's just because of pragma Unreferenced, it is trivial to patch it to
>> > compile with 3.15p.
>> 
>> The problem I found with Florist, Glade and ASIS is that most ACT Tools
>> compile with -gnatg which implies -gnatwa, -gnatwe and-gnaty.
>> 
>> A fix is easy but time consuming.
> 
> I do not understand what the problem is.  All three packages build
> just fine on Debian with 3.15p.  By checking the buildd logs, I
> have gathered the flags used:
> 
> ASIS static lib:  -g -O2 -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
> ASIS shared lib:  -O2 -fPIC -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
> ASIS tools:       -gnaty -gnatg -gnatwu -gnatwe -gnato -gnatn -gnatVa
> Florist static:   -g -O2 -gnatpg
> Florist shared:   -O2 -fPIC -gnatpg
> Glade:            -gnatpg -O2 -gnatpn -gnatwu -gnatwe
> 
> Are you trying to say that GCC 3.4 does not support some of these flags?

No, gcc 3.4 has more warnings. For example one warning which detects
variables which could be decalred constant but are not.  -gnatwa activates
them. The old sources have hundreds of places where the new warnings are
triggered.

With Regards

Martin

-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: PolyORB 1.1r

[Ludovic Brenta]

Martin Krischik wrote:
> No, gcc 3.4 has more warnings. For example one warning which detects
> variables which could be decalred constant but are not.  -gnatwa activates
> them. The old sources have hundreds of places where the new warnings are
> triggered.

Ah, OK.  This is a good thing.  GNAT has a long history of becoming ever
more strict.

Now, this also means that it should be easier to use GNAT 3.15p than
GCC 3.4 when compiling PolyORB for Debian.

BTW, GCC 3.4 is not in Debian "unstable" yet, because it breaks ABI
compatibility on some architectures.  It is in Debian's "experimental"
distribution.

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Jerome Hugues]

In article <87y8mx8g4p.fsf@insalien.org>, Ludovic Brenta wrote:

> Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
> it's just because of pragma Unreferenced, it is trivial to patch it to
> compile with 3.15p.

PolyORB uses no Ada 2005 features. PolyORB may compile with 3.15p,
there are still work around for 3.15 concerning e.g. pragma Unreferenced.

PolyORB does not require GCC 3.4.0, it is just that it is no more
tested with 3.15p. Besides, we got feedback that it compiles with
3.15p on Linux.

-- 
Jerome

Re: PolyORB 1.1r

[Ludovic Brenta]

Jerome Hugues <hugues@nospam*merlin.enst.fr> wrote in message news:<slrnccdga9.41c.hugues@merlin.enst.fr>...
> In article <87y8mx8g4p.fsf@insalien.org>, Ludovic Brenta wrote:
> 
> > Why does PolyORB require GCC 3.4? Does it use Ada 2005 features?  If
> > it's just because of pragma Unreferenced, it is trivial to patch it to
> > compile with 3.15p.
> 
> PolyORB uses no Ada 2005 features. PolyORB may compile with 3.15p,
> there are still work around for 3.15 concerning e.g. pragma Unreferenced.
> 
> PolyORB does not require GCC 3.4.0, it is just that it is no more
> tested with 3.15p. Besides, we got feedback that it compiles with
> 3.15p on Linux.

That is what I thought.  So, Bruce, there is nothing stopping you from
packaging PolyORB for Debian.  The worst thing you may have to do is
sed 's/(pragma +Unreferenced)/--\1/g' all the source files.

Although I do not have the time to do it myself, feel free to ask me
questions and I'll try to help.  You may want to peruse my Debian
Policy for Ada[1].

[1] http://users.skynet.be/ludovic.brenta/debian-ada-policy.html

-- 
Ludovic Brenta.

Re: PolyORB 1.1r

[Dr. Adrian Wrigley]

...
> Other instantiations of PolyORB are available in the public PolyORB
> CVS repository for testing purposes. Available instantiations include
> DSA (Distributed System Annex), MOMA (Message Oriented Middleware for
> Ada) and AWS (Ada Web Server).
...

A few questions come to mind...

Can I take an existing Annex E/Glade program and recompile it
with PolyORB? (no modifications needed?)  Is there a porting guide?

Is this intended to supersede Glade?  Does it do everything
Glade could do, only better?  Is PolyORB/DSA suitable in
a "production" system where high reliability is wanted?

How does it compare with Glade in terms of speed, complexity, code size,
robustness etc?

I currently use Glade, but have been unable to fix a couple of
problems with the implementation, and there don't seem to be
any versions beyond 3.15 on the horizon.

(I know I could probably answer these questions myself in a few
hours or days of experimentation, but you guys are much better
placed to explain the pros and cons of the technologies.)

Thanks.
-- 
Dr Adrian Wrigley, Cambridge, UK

Re: PolyORB 1.1r

[Martin Krischik]

Dr. Adrian Wrigley wrote:

> Is this intended to supersede Glade?  Does it do everything

> I currently use Glade, but have been unable to fix a couple of
> problems with the implementation, and there don't seem to be
> any versions beyond 3.15 on the horizon.

I have tied to port Glade to gcc 3.5 and failed. gcc 3.5 does not implement
the -gnatz options any more. They are still there but won't create any .o
files any more.

I take this an an indication that Glade is indeed dead.

With Regards

Martin
-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: PolyORB 1.1r

[Laurent Pautet]

Martin Krischik <krischik@users.sourceforge.net> writes:

> Dr. Adrian Wrigley wrote:
>
> I take this an an indication that Glade is indeed dead.

Let's summarize the story.

1995
   GLADE first and unique ORB for DSA (Distributed Systems Annex) done
   by a research team (mostly L. Pautet and S. Tardieu) at T�l�com
   Paris with the help of CSC and then supported by ACT (at this time).

1999
   AdaBroker a free ORB for CORBA done by the same research team at
   T�l�com Paris (basically GLADE team, T. Quinot and other
   contributors).

2002
   PolyORB a generic ORB (which provides DSA, CORBA, MOM, ...) has
   been designed by a research team from T�l�com Paris, LIP6 and ACTE
   (mostly L. Pautet, T. Quinot, J. Hugues and F. Kordon). This
   middleware is based on GLADE and AdaBroker know-how. PolyORB is
   supported by AdaCore. It is also the result of the current research
   work on middleware architecture carried out by T�l�com Paris/LIP6.
   Many papers in non-Ada conferences.

GLADE is definitively alive. There are customer releases for 3.16 and
5.02. Unfortunately, there is no public release from a long time. The
GLADE CVS tree may soon become available. Note that when GLADE is
integrated in PolyORB, GLADE will not be dead. It will be part of a
larger middleware. GNAT will probably still be the only Ada
environment providing Annex E.

PolyORB is definitively alive. It is a successful professional ORB and
more important it is a successful Ada research project in the free
software area and in the middleware community (many papers in non-Ada
conferences).

I can tell you it is a hard job to be part of the middleware community
when you are using Ada, when you have to deal with C++ ORBs (even in
the RT community) like TAO (intensively supported through DARPA
projects) and when you are not part of the Java reflective bla
approach.

So yes, we don't have a lot of time to communicate about GLADE,
AdaBroker and PolyORB on the web or through press releases. But this
is not really our job :) But well, we do our best to use Ada to do our
research work. We communicate on it in conferences (therefore we
promote Ada mostly outside the Ada community).
 
--
--  Laurent

Re: PolyORB 1.1r

[Martin Krischik]

Laurent Pautet wrote:

> Martin Krischik <krischik@users.sourceforge.net> writes:
 
>> Dr. Adrian Wrigley wrote:
>>
>> I take this an an indication that Glade is indeed dead.

> So yes, we don't have a lot of time to communicate about GLADE,
> AdaBroker and PolyORB on the web or through press releases. But this
> is not really our job :) But well, we do our best to use Ada to do our
> research work. We communicate on it in conferences (therefore we
> promote Ada mostly outside the Ada community).

I just had the impression that PolyORB would fully replace GLADE. I never
meant to criticise you excellent effort.

With Regards

Martin

-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: PolyORB 1.1r

[Laurent Pautet]

"Dr. Adrian Wrigley" <amtw@linuxchip.demon.co.uk.uk.uk> writes:

> Can I take an existing Annex E/Glade program and recompile it
> with PolyORB? (no modifications needed?)  Is there a porting guide?

The answer is "yes" an existing Annex E/Glade program works with
GNAT/DSA/PolyORB (we pass ACATS for instance) and "no" you cannot port
it yourself because you need the patch to GNAT to produce code for
PolyORB (and not for GARLIC which is the GLADE ORB).

> Is this intended to supersede Glade?  Does it do everything
> Glade could do, only better?  Is PolyORB/DSA suitable in
> a "production" system where high reliability is wanted?

Yes. It may be integrated in 5.03. Not sure about that.

> How does it compare with Glade in terms of speed, complexity, code size,
> robustness etc?

Almost the same with much more features. For instance, you can make a
CORBA program and a DSA program interoperate.

Note that PolyORB is also a research project. That is why there are
public releases (p) and research releases (r). The customer and public
releases are intensively tested on many platforms. There is a strong
commitment from AdaCore to support these releases. A research release
is a stable snapshot tested on a bunch of platforms. Its main goal is
to help people to contribute to this research work. But AdaCore does
not support these specific releases (as the testing activity is much
lighter).

> I currently use Glade, but have been unable to fix a couple of
> problems with the implementation, and there don't seem to be
> any versions beyond 3.15 on the horizon.

This is another point. There has been no public release of GLADE for a
long time. But there are professional releases for 3.16 and 5.02.

The good news is that the GLADE CVS tree may become soon available. I
hope to have a GO before Ada-Europe 2004. So very soon ... Stay tuned!

Thank you for patience,

--
--  Laurent