gcc-3.2 build problem in the java area

gcc-3.2 build problem in the java area

[cyan]

Hi,

I'm having consistent problem compiling gcc-3.2 on a Solaris 7 box.
The problem seems to be with libjava/java/net/natInetAddress.cc and
the make bombs out with:

./../../../libjava/java/net/natInetAddress.cc: In static member function 
`static java::lang::String* java::net::InetAddress::getLocalHostname()':
../../../../libjava/java/net/natInetAddress.cc:338: `MAXHOSTNAMELEN'
undeclared 
   (first use this function)
../../../../libjava/java/net/natInetAddress.cc:339: `buffer' undeclared
(first use this function)
make[5]: *** [java/net/natInetAddress.lo] Error 1

I'm compiling on an E420R, Solaris 7, kernel Generic_106541-21. I've
tried various things such as trying the build with binutils-2.13, but
it's the same error that occurs.

I noticed in your archives, back in June 1, 2002, a gentleman named
George Goffe had almost the exact same problem I'm having, except
he was trying to compile gcc-3.1 on a Solaris 8 box:

http://gcc.gnu.org/ml/java/2002-06/msg00002.html

Any pointers will be appreciated.

Thanks in advance!

Charlie

----------------------------------------------------------------------
Charlie Yanaitis                        charles.yanaitis@rochester.edu
Systems Analyst/Programmer                              (585) 275-4189
Information Technology Services                University of Rochester
----------------------------------------------------------------------

gcc-3.2 build problem in the java area

[Andrew Haley]

cyan@obrien.cc.rochester.edu writes:
 > Hi,
 > 
 > I'm having consistent problem compiling gcc-3.2 on a Solaris 7 box.
 > The problem seems to be with libjava/java/net/natInetAddress.cc and
 > the make bombs out with:
 > 
 > ./../../../libjava/java/net/natInetAddress.cc: In static member function 
 > `static java::lang::String* java::net::InetAddress::getLocalHostname()':
 > ../../../../libjava/java/net/natInetAddress.cc:338: `MAXHOSTNAMELEN'
 > undeclared 
 >    (first use this function)
 > ../../../../libjava/java/net/natInetAddress.cc:339: `buffer' undeclared
 > (first use this function)
 > make[5]: *** [java/net/natInetAddress.lo] Error 1
 > 
 > I'm compiling on an E420R, Solaris 7, kernel Generic_106541-21.

You should find this in /usr/include/netdb.h:

#define    MAXHOSTNAMELEN  256

Is it there?

Andrew.

Re: gcc-3.2 build problem in the java area

[cyan]

Andrew

On Thu, 10 Oct 2002, Andrew Haley wrote:

> cyan@obrien.cc.rochester.edu writes:
>  > Hi,
>  > 
>  > I'm having consistent problem compiling gcc-3.2 on a Solaris 7 box.
>  > The problem seems to be with libjava/java/net/natInetAddress.cc and
>  > the make bombs out with:
>  > 
>  > ./../../../libjava/java/net/natInetAddress.cc: In static member function 
>  > `static java::lang::String* java::net::InetAddress::getLocalHostname()':
>  > ../../../../libjava/java/net/natInetAddress.cc:338: `MAXHOSTNAMELEN'
>  > undeclared 
>  >    (first use this function)
>  > ../../../../libjava/java/net/natInetAddress.cc:339: `buffer' undeclared
>  > (first use this function)
>  > make[5]: *** [java/net/natInetAddress.lo] Error 1
>  > 
>  > I'm compiling on an E420R, Solaris 7, kernel Generic_106541-21.
> 
> You should find this in /usr/include/netdb.h:
> 
> #define    MAXHOSTNAMELEN  256
> 
> Is it there?
> 
> Andrew.

Yes, that is one of the things I did check. How to I get the make
to recognize this? I tried fooling it by by setting CFLAGS to
-DMAXHOSTNAMELEN=256, but the make still bombed out in the same
place.

Thanks!

----------------------------------------------------------------------
Charlie Yanaitis                        charles.yanaitis@rochester.edu
Systems Analyst/Programmer                              (585) 275-4189
Information Technology Services                University of Rochester
----------------------------------------------------------------------

Re: gcc-3.2 build problem in the java area

[Russ Allbery]

Andrew Haley <aph@redhat.com> writes:
> cyan@obrien.cc.rochester.edu writes:

>> ./../../../libjava/java/net/natInetAddress.cc: In static member function 
>> `static java::lang::String* java::net::InetAddress::getLocalHostname()':
>> ../../../../libjava/java/net/natInetAddress.cc:338: `MAXHOSTNAMELEN'
>> undeclared 
>>    (first use this function)
>> ../../../../libjava/java/net/natInetAddress.cc:339: `buffer' undeclared
>> (first use this function)
>> make[5]: *** [java/net/natInetAddress.lo] Error 1
>> 
>> I'm compiling on an E420R, Solaris 7, kernel Generic_106541-21.

> You should find this in /usr/include/netdb.h:

> #define    MAXHOSTNAMELEN  256

> Is it there?

Solaris only defines MAXHOSTNAMELEN if:

#if !defined(_XPG4_2) || defined(__EXTENSIONS__)

Since this is C++, I'm guessing that it's including the headers in a more
strictly standard-conforming mode, which means that this #define is
omitted.

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>

Re: gcc-3.2 build problem in the java area

[Tom Tromey]

>>>>> "Russ" == Russ Allbery <rra@stanford.edu> writes:

Russ> Solaris only defines MAXHOSTNAMELEN if:
Russ> #if !defined(_XPG4_2) || defined(__EXTENSIONS__)

Is there any drawback to simply always defining __EXTENSIONS__?
We could do that in include/posix.h.

Tom

Re: gcc-3.2 build problem in the java area

[Andrew Haley]

Tom Tromey writes:
 > >>>>> "Russ" == Russ Allbery <rra@stanford.edu> writes:
 > 
 > Russ> Solaris only defines MAXHOSTNAMELEN if:
 > Russ> #if !defined(_XPG4_2) || defined(__EXTENSIONS__)
 > 
 > Is there any drawback to simply always defining __EXTENSIONS__?
 > We could do that in include/posix.h.

But who might know what else __EXTENSIONS__ might do?  I suppose we
could "#ifdef __solaris" or somesuch.

It might be lees risky to "#ifdef __solaris .. #define MAXHOSTNAMELEN 256

Andrew.

Re: gcc-3.2 build problem in the java area

[David Edelsohn]

>>>>> Andrew Haley writes:

Andrew> But who might know what else __EXTENSIONS__ might do?  I suppose we
Andrew> could "#ifdef __solaris" or somesuch.

	Maybe __EXTENSIONS__ needs to be added to CPLUSPLUS_CPP_SPEC for
that target?

David

Re: gcc-3.2 build problem in the java area

[Tom Tromey]

>>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:

Andrew> It might be lees risky to "#ifdef __solaris .. #define
Andrew> MAXHOSTNAMELEN 256

Or #ifndef MAXHOSTNAMELEN ...

Tom

Re: gcc-3.2 build problem in the java area

[Andrew Haley]

Tom Tromey writes:
 > >>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:
 > 
 > Andrew> It might be lees risky to "#ifdef __solaris .. #define
 > Andrew> MAXHOSTNAMELEN 256
 > 
 > Or #ifndef MAXHOSTNAMELEN ...

Hell, whatever works gets my vote!  All we need is a nice person out
there with a Solaris system to submit a patch...

Andrew.

Re: gcc-3.2 build problem in the java area

[Tom Tromey]

>>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:

Andrew> Hell, whatever works gets my vote!  All we need is a nice person out
Andrew> there with a Solaris system to submit a patch...

Can the original reporter try this?
Andrew, if you know who the original reporter is, could you forward it?

Tom

Index: ChangeLog
from  Tom Tromey  <tromey@redhat.com>

	* java/net/natInetAddress.cc (MAXHOSTNAMELEN) [!WIN32]: Define if
	not defined.

Index: java/net/natInetAddress.cc
===================================================================
RCS file: /cvs/gcc/gcc/libjava/java/net/natInetAddress.cc,v
retrieving revision 1.19
diff -u -r1.19 natInetAddress.cc
--- java/net/natInetAddress.cc 28 Feb 2002 01:03:37 -0000 1.19
+++ java/net/natInetAddress.cc 11 Oct 2002 18:04:13 -0000
@@ -1,6 +1,6 @@
 // natInetAddress.cc
 
-/* Copyright (C) 1998, 1999, 2000  Free Software Foundation
+/* Copyright (C) 1998, 1999, 2000, 2002  Free Software Foundation
 
    This file is part of libgcj.
 
@@ -42,6 +42,10 @@
 #ifdef HAVE_NETDB_H
 #include <netdb.h>
 #endif
+
+#ifndef MAXHOSTNAMELEN
+#define MAXHOSTNAMELEN	256
+#endif /* MAXHOSTNAMELEN */
 
 #endif /* WIN32 */
 

Re: gcc-3.2 build problem in the java area

[Andrew Haley]

Tom Tromey writes:
 > >>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:
 > 
 > Andrew> Hell, whatever works gets my vote!  All we need is a nice person out
 > Andrew> there with a Solaris system to submit a patch...
 > 
 > Can the original reporter try this?
 > Andrew, if you know who the original reporter is, could you forward it?

Done.

Andrew.

Re: gcc-3.2 build problem in the java area

[Kaveh R. Ghazi]

 > Solaris only defines MAXHOSTNAMELEN if:
 > 
 > #if !defined(_XPG4_2) || defined(__EXTENSIONS__)
 > 
 > Since this is C++, I'm guessing that it's including the headers in
 > a more strictly standard-conforming mode, which means that this
 > #define is omitted.

... or it means that _XPG4_2 is defined.  Or something else is wrong.
We should understand what this is first.

I've been testing solaris2 daily (forever it seems) and I've never had
this problem.  Heck, there's an autobuild bootstrapping 3.2 on
solaris2 and posted daily too.  Look at the gcc-testsuite results
list.  Before we start installing patches we should figure out what
this person did differently so we can find the correct remedy.  It may
be something properly addressed by a different patch or by fixing the
installation docs or ... ?  But IMHO we need to know why this one
person had the problem and numerous others don't.

		Thanks,
		--Kaveh

--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu

Re: gcc-3.2 build problem in the java area

[Jeff Sturm]

On Fri, 11 Oct 2002, Andrew Haley wrote:
> Hell, whatever works gets my vote!  All we need is a nice person out
> there with a Solaris system to submit a patch...

I have Solaris, and I have never seen this failure.  Though I haven't done
a Solaris build recently.

Trying now... should know by tomorrow :(

Jeff