Sorry Tom, didnt realize that 'Reply' to the mail sends it only to you.<div><br></div><div>Regarding your reply , I do not have any 'Long double' in my IDL file, I have only 'long' . The error message suggests that the undefined 'long double' are from the files in the INCLUDE directory, not in files generated from my idl </div>
<div><br></div><div>Is there something wrong in my compilation (Makefile below) ? </div><div><br></div><div><br></div><div><div>CC = /usr/bin/c++</div><div>CPPFLAGS = -g -c</div><div>LDFLAGS = -g</div>
<div>OMNI_HOME = /filer_build/ssalvady/buildreleases/usr/omniORB-4.0.5</div><div>OMNI_INCLUDES = -I$(OMNI_HOME)/include</div><div>OMNI_LIB_DIR = $(OMNI_HOME)/lib</div><div>OMNIIDL = $(OMNI_HOME)/bin/omniidl</div>
<div>INCLUDES = $(OMNI_INCLUDES)</div><div>LIBS = -lomniORB4 -lomnithread -lomniDynamic4</div><div>OBJECTS = channelAttributes.o channelAttributesImpl.o server.o</div><div><br></div><div>all: $(OBJECTS)</div>
<div><span class="Apple-tab-span" style="white-space:pre"> </span>$(CC) $(LDFLAGS) -o server -L$(OMNI_HOME)/lib $(OBJECTS) $(LIBPATH) $(LIBS)</div><div><br></div><div>channelAttributes.o: channelAttributes.cxx channelAttributes.h</div>
<div><span class="Apple-tab-span" style="white-space:pre"> </span>$(CC) $(CPPFLAGS) $(INCLUDES) -L$(OMNI_HOME)/lib channelAttributes.cxx</div><div><br></div><div>server.o: server.cc channelAttributes.h</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>$(CC) $(CPPFLAGS) $(INCLUDES) server.cc</div>
<div><br></div><div>channelAttributesImpl.o: channelAttributesImpl.cc channelAttributesImpl.h channelAttributes.h</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>$(CC) $(CPPFLAGS) $(INCLUDES) channelAttributesImpl.cc</div>
<div><br></div><div>channelAttributes.cxx: channelAttributes.idl</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>$(OMNI_HOME)/bin/omniidl -bcxx -Wbh=.h channelAttributes.idl</div><div><br></div><div>
<br></div>Thanks </div><div>Sankar</div><div><br></div><div><br></div><div><br></div><div><div class="gmail_quote">On Tue, Oct 19, 2010 at 6:37 PM, Thomas Lockhart <span dir="ltr"><<a href="mailto:lockhart@fourpalms.org">lockhart@fourpalms.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div text="#000000" bgcolor="#ffffff">
Please always reply to the list.<div class="im"><br>
<span style="white-space:pre-wrap"><br>
> I tried c++ instead of g++ , the result is the same.<br>
> How do I fix the configuration option ? Any quick pointers
would be<br>
> great, Tom.<br>
</span><br></div>
I do not use the long double in my code, but it seems to be
supported in my RPM build of omniORB-4.1.4 on Fedora 13. You are
using very old versions of everything and I don't have those
versions available to me to look at.<br>
<br>
I do see in this newer version that "long double" in an IDL file
produces references to CORBA::LongDouble in the stubs and skeletons.
So that part seems OK.<br>
<br>
In this newer version, definitions are guarded by #if HAS_LongDouble
so you may want to look at the omniORB header files for that and see
why it might not be defined. I would think that you will need it
defined when building omniORB to have it properly available when
trying to use it. otoh I would have guessed that the omniidl
compiler would have not produced code referencing CORBA::LongDouble
if it was disabled during the package build.<br>
<br>
hth<br>
<br>
- Tom<br>
</div>
</blockquote></div><br></div>