Re: [mico-devel] Why is CosEventComm::PushConsumer_skel no longer generated in 2.3.13?

2009-01-16 Thread Andreas Benzler
Hi Klaus

I am highly interested in your approach bringing mico 2.3.11 to a gcc4.3
platform. Can you please describe a bit more detail what you did
exactly. I am not an automake expert :-(.
If you could make a patch available for 2.3.11 it would be even better...

Regards

Andreas Benzler


Kretschel Klaus schrieb:
 Hi Karel,
 
 On Wednesday 14 January 2009 15:41, Karel Gardas wrote:
 exactly! MICO 2.3.13 does not support BOA anymore. For your information,
 your CosEventComm::PushConsumer_skel is BOA's skeleton code. My best
 advice to you is to update code to use POA's skeletons. Hmm, why do you
 need to use skeleton code exactly? It's used just by service
 implementors, isn't it? Or are you writing your own event service? If
 so, then ok...
 
 it's old code written by someone else, and I guess it is not necessary. Can 
 you recommend some useful documentation for POA event service? The Puder book 
 does not even mention the event service...
 
 But we will not need to upgrade by now. Using the configure scripts from 
 2.3.13 allowed also to compile 2.3.11 with gcc 4.3. The key was 
 the -fno-strict-aliasing option.
 
 Best regards,
 Klaus
 ___
 Mico-devel mailing list
 Mico-devel@mico.org
 http://www.mico.org/mailman/listinfo/mico-devel
 
___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel


Re: [mico-devel] How to compile mico with externalize enabled?

2009-01-12 Thread Andreas Benzler
Hi Karel

Thank for your fast info.
Currently my time is very limited and I do not work on mico stuff in the
first place, so I am sorry but I probably won't have the time to help
out here.
The reason I looked at the externalize service was that we encountered
problems to serialize mico objects in the server when a QT application
is the client. (See my other posts on the list). I am not an mico /
CORBA expert but I thought externalizing is a service that does
serializing as well and I just wanted to give it a short try...
I have a student working on that serializing problem. I just wanted to
be able to give him some more hints to look for because he still could
not solve the problem.

Sorry that I currently can not help out.

Andreas
___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel


[mico-devel] How to compile mico with externalize enabled?

2009-01-09 Thread Andreas Benzler
Hi all

I would like to compile mico 2.3.13 with externalize support but all I
got was a lot of errors. I tried different configurations to solve the
problem but nothing worked out. Does some of you could manage to compile
mico with externalize support?

Here is what I did:

first try:
./configure --enable-externalize

resulting compile errors:
In file included from CosExternalization.cc:8:
./CosExternalization.h:123:31: error: coss/CosLifeCycle.h: No such file
or directory
./CosExternalization.h:124:36: error: coss/CosObjectIdentity.h: No such
file or directory
./CosExternalization.h:127:35: error: coss/CosRelationships.h: No such
file or directory
./CosExternalization.h:128:28: error: coss/CosGraphs.h: No such file or
directory

next try:
./configure --enable-externalize --enable-life

resulting compile errors:
./CosCompoundLifeCycle.h:64:36: error: coss/CosObjectIdentity.h: No such
file or directory
./CosCompoundLifeCycle.h:67:35: error: coss/CosRelationships.h: No such
file or directory
./CosCompoundLifeCycle.h:68:28: error: coss/CosGraphs.h: No such file or
directory

next try:
./configure --enable-externalize --enable-life --enable-relship

resulting compile errors:
FactoryFinder_impl.cc:27:34: error: coss/CosTradingRepos.h: No such file
or directory
In file included from ../../include/coss/FactoryFinder_impl.h:30,

 from FactoryFinder_impl.cc:28:

../../include/coss/RegisterHelper.h:3:29: error: coss/CosTrading.h: No
such file or directory
In file included from FactoryFinder_impl.cc:28:

../../include/coss/FactoryFinder_impl.h:35: error: ‘CosTrading’ has not
been declared


next try:
 ./configure --enable-externalize --enable-life --enable-relship
--enable-trader

resulting compile errors:
/X/mico/./idl/idl --no-poa --boa --c++-skel -B../.. --any --name
CosTrading ../../include/coss/CosTrading.idl
error: --boa option is no longer supported
   (if you need BOA support, please use MICO release older than 2.3.12)




I am using a SUSE 11.1 with gcc 4.3 and the mico 2.3.13 release.

Thanks for helping

Andreas

___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel


[mico-devel] Problems using Objects of type Any and Qt

2008-12-03 Thread Andreas Benzler
Hallo,
I´m using a client-server model based on Mico and I defined an idl
interface like this:

interface ifRec
{
long Register( in string name, in string ns );
void WriteData( in long id, in any data );
void SetEnv( in string Key, in string Value );
void Rec();
void Break();
void Resume();
void Replay( in double speed );
void Stop();
long Load( in string filename );
};

The WriteData function uses the parameter data of type Any, which
represents a structure (struct) containing other structures inside, and I
get a run-time error whenever a client calls this function. The client has
been compiled with Qt and uses the QApplication function. I discovered
that the run-time error it´s based on the problem that the object of type
Any is codified in the wrong way from Mico during the serialization
process only when the QApplication function is used in the client code and
also when the tree´s depth of the structures codified in the Any object is
too big, otherwise no run-time error comes up and the octet sequence that
represents the object, made by the Mico serialization process, has the
correct length. I also discovered that the difference between the octet
sequence with and without using QApplication is fixed and always 268 bytes
(more bytes without QApplication) and it does not depend on the complexity
of the structure that the Any object represents. I discovered this last
detail simulating a serialization process of the Any object using the
available Mico internal codec. I tried to compile the system with the two
Mico versions 2.3.11 and 2.3.13, both versions compiled specifying also the
--with-qt=qt-path directive for the configure script, but I have same
problems.

Here is also the Mico codec I used to simulate the serialization process:

void operator ( const Any anyobject, OctetSeq dataseq)
{
// Codec
ORB_var orb;
Object_var obj;
IOP::CodecFactory_var cf;
IOP::Encoding enc2;
IOP::Codec_ptr codec_ptr;

// Generate codec based on CodecFactory
int argc = 1;
char* argv[1];
argv[0] = anyserial;
orb = ORB_init( argc, argv, mico-local-orb );
obj = orb-resolve_initial_references(CodecFactory);
cf = IOP::CodecFactory::_narrow(obj);
enc2.format = IOP::ENCODING_CDR_ENCAPS;
enc2.major_version = 1;
enc2.minor_version = 2;
codec_ptr = cf-create_codec(enc2);

// Encode anyobject and generate a new octet sequence
OctetSeq_var newdataseq_ptr = codec_ptr-encode(anyobject);
int length = newdataseq_ptr-length();

// Copy generated octet sequence into dataseq
dataseq.length(length);

for (int i=0; ilength; i++)
dataseq[i] = (*newdataseq_ptr)[i];
}

Any and OctetSeq types are exactly CORBA::Any and CORBA::OctetSeq types
defined in MICO.

Could please anyone help to understand why Mico has this strange behavior
and eventually to find a solution?

Many Thanks in advance,
Pasquale Zarcone

Because the mail did not come trough with my colleges account I post
this message on his behalf

Andreas
___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel


Re: [mico-devel] Problems compiling mico

2008-11-07 Thread Andreas Benzler
I have to correct my last post a bit.
As I wrote I installed MICO 2.3.13 on Kubuntu 8.10 without problems I
did not mention that I had to apply the patch that was posted on this
list to avoid the error 'fast_array.cc:51: error: 'UINT_MAX' was not
declared in this scope'
I had a look about the gcc version on Kubuntu and it states:
gcc (Ubuntu 4.3.2-lubuntull) 4.3.2

But the gcc team seems to change some of the headers.
I compiled MICO 2.3.13 without any patches and problems on a SUSE 11.1
beta4. The gcc --version output here is:
gcc (SUSE Linux) 4.3.2 [gcc-4_3-branch revision 141291]


Andreas

___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel


Re: [mico-devel] Problems compiling mico

2008-11-06 Thread Andreas Benzler
Hello Guilherme 

It might be a gcc 4.3 problem. I have not looked at the code cuasing your
problem, code but here http://gcc.gnu.org/gcc-4.3/porting_to.html is this
error mentioned because some header includes were removed in the new gcc.
There is stated the the header cstring has to be included to solve the
problem.
By the way, I compiled mico 2.3.13 a couple of days ago myself on a kubuntu
8.10 without any problems, but I hav not checked which gcc version it was
exactly.


Andreas Benzler

   In file included from ../include/CORBA.h:280,
from ../include/CORBA-SMALL.h:52,
from os-unix.cc:34:
   ../include/mico/throw.h: In function 'void
mico_sii_throw(CORBA::StaticRequest*, ...)':
   ../include/mico/throw.h:111: error: 'strcmp' was not declared in
this scope

___
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel