[omniORB] Build Problem on HPUX 11.00 with aCC(A.03.25)

[Paul Nader]

Sai-Lai,

Are the changes Masaaki suggested likely to make their may into the development
branch?

Thanks, Paul.

Masaaki Sekiya wrote:

> Hello, omniORBers.
>
> I get compile Error and Warnings with HP aCC (B3910B A.03.25).
> To remove it , I apply patches to some files.
>
> My question is:
> Are there something wrong in these patches ?
> Is it possible to apply these ( or some of ) patches to external omniORB3.0.1 ?
>
> * Error caused by compiler bug (maybe)
>
> I have used omniORB_3.0.1 with HP aCC which version is as below.
>
> /usr/bin/aCC:
>         HP aC++ B3910B A.03.13
>         HP aC++ B3910B X.03.11.10 Language Support Library
>
> Then, I changed aCC version to as below.
>
> /usr/bin/aCC:
>         HP aC++ B3910B A.03.25
>         HP aC++ B3910B X.03.14.12 Language Support Library
>
> =============
> aCC -c -O -w +inst_v +DAportable -D_THREAD_SAFE -DRWSTD_MULTI_THREAD -DRW_MULTI_
> THREAD  -DIDLMODULE_VERSION="\"0x2301\"" -I/usr/local/python/include +Z -I.  -I.
> ./../../../include -D__hppa__ -D__hpux__ -D__OSVERSION__=11 -ldir y.tab.cc -o y.
> tab.o
> Error 439: Exact position unknown; near ["./idlscope.h", line 244]. # Name 'Kind
> ' was redefined after its use in a class. A previous declaration used "enum Scop
> e::Kind" ["./idlscope.h", line 119].
> =============
> This error is issued before ,
> http://www.uk.research.att.com/omniORB/archives/2000-09/0022.html
>
> As Paul wrote , it looks like a compiler bug , which is related with
> scope-identifier collision.
> To solve it, patch like below may be necessary.
> ( change enum Kind to oter name )
>
> % diff idlscope.h
> 244c244
> <     enum EKind {
> ---
> >     enum Kind {
> 254c254
> <     Entry(const Scope* container, EKind kind, const char* identifier,
> ---
> >     Entry(const Scope* container, Kind kind, const char* identifier,
> 261c261
> <     EKind              kind()       const { return kind_; }
> ---
> >     Kind              kind()       const { return kind_; }
> 279c279
> <     EKind              kind_;
> ---
> >     Kind              kind_;
> % diff idlscope.cc
> 216c216
> < Entry(const Scope* container, EKind k, const char* identifier,
> ---
> > Entry(const Scope* container, Kind k, const char* identifier,
>
> ( To change place of a declaration of enum also may solve error )
>
> And I want to remove some warnings.
>
> * omkdepend cannot open "+Z"
> =============
> ../../../../bin/hppa_hpux_11.00/omkdepend -D__cplusplus -DIDLMODULE_VERSION="0x2
> 301" -I/usr/local/python/include +Z -I. -I../../../../include -D__hppa__ -D__hpu
> x__ -D__OSVERSION__=11 idlc.cc idlpython.cc idlconfig.cc idldump.cc idlvalidate.
> cc idlast.cc idlexpr.cc idlscope.cc idlrepoId.cc idltype.cc idlutil.cc idlerr.cc
>  lex.yy.cc y.tab.cc
> ../../../../bin/hppa_hpux_11.00/omkdepend: warning:  cannot open "+Z"
> =============
> This warning may come from omkdepend bug.
> Patch for main.c may be necessary.
>
> %diff main.c
> 212c212
> <                       if (endmarker && **argv == '+')
> ---
> >                       if (endmarker || **argv == '+')
>
> In order to check strictly,
> I add complile option "+w" to CXXOPTIONS in hppa_hpux_11.00.mk.
> Then I get some future error and suggestion from compiler.
>
> * future Error because of union without name
> =============
> Error (future) 254: "./typecode.h", line 158 # Anonymous unions may only contain
>  non-static data members and not define types.
> =============
> To remove it, I add name of unions.
>
> * explicit private
> =============
> Warning (suggestion) 555: "dynAnyNil.cc", line 206 # Access type should be speci
> fied in the base class list; the default is private.
> =============
> To remove it , I add 'private' word.
>
> * casting 'sockaddr_in *' -> 'struct sockaddr *'
> This warning appears in 'tcpSocketMTfactory.cc'
> To remove it, I apply patche using 'reinterpret_cast' as below.
>
> < (struct sockaddr *)&raddr
> ---
> > reinterpret_cast<sockaddr*>(&raddr)
>
> And next two cast looks internal matter.
> So I can't see weather a patch just using 'reinterpret_cast' is enough or not.
>
> * casting 'omniOrbPOAManager *' -> 'CORBA::Object_ptr'/'PortableServer::POA_ptr'
> =============
> Warning 749: "poamanager.cc", line 294 # The cast from 'omniOrbPOAManager *' to
> PortableServer::POA *' is performed as a 'reinterpret_cast'. This operation is
> non-portable and potentially unsafe.
>         return (PortableServer::POA_ptr) this;
> =============
>
> * casting 'CORBA::TypeCode_ptr' -> 'TypeCode_base *'
> =============
> Warning 749: "../poamanager.cc", line 294 # The cast from 'omniOrbPOAManager *'
> to 'PortableServer::POA *' is performed as a 'reinterpret_cast'. This operation
> is non-portable and potentially unsafe.
>         return (PortableServer::POA_ptr) this;
> =============
>
> Thanks in advance.
>
> Masaaki Sekiya (YDC)