From: Greg Larkin on
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Gerald Pfeifer wrote:
> On Fri, 9 Apr 2010, Gerald Pfeifer wrote:
>> As for the lang/gcc42 issue, I have an idea of how to fix that and will
>> give it try now and share the patch if it works for me.
>
> The patch below passes testing for me, and I consider it The Right
> Thing[TM] directionally, too. The two actual changes are the
> addition of --with-as=/usr/bin/as and --with-ld=/usr/bin/ld.
>
> I have not committed it yet, but am planning on doing so and
> definitely am interested in whether this changes (improves) things
> for you.
>
> That said, please do not stop on moving from GCC 4.2 to GCC 4.4 when
> it comes to Java, even in case this patch works for you. :-)
>
> Gerald
>
> Index: Makefile
> ===================================================================
> RCS file: /home/pcvs/ports/lang/gcc42/Makefile,v
> retrieving revision 1.356
> diff -r1.356 Makefile
> 56,60c56
> < CONFIGURE_ARGS+= --disable-nls \
> < --with-system-zlib \
> < --with-libiconv-prefix=${LOCALBASE} \
> < --with-gmp=${LOCALBASE} \
> < --program-suffix=${SUFFIX} \
> ---
>> CONFIGURE_ARGS+=--disable-nls \
> 63c59,65
> < --with-gxx-include-dir=${TARGLIB}/include/c++/
> ---
>> --program-suffix=${SUFFIX} \
>> --with-as=/usr/bin/as \
>> --with-gmp=${LOCALBASE} \
>> --with-gxx-include-dir=${TARGLIB}/include/c++/ \
>> --with-ld=/usr/bin/ld \
>> --with-libiconv-prefix=${LOCALBASE} \
>> --with-system-zlib

Thank you Gerald, I'll give that a try with gcc 4.2 and gcc 4.5
installed. I think we're getting close with gcj 4.5, assuming my fix
for duplicate Java symbols is acceptable.

I need to do some more tinderbox builds and testing and clear up some
other errors reported by users, but maybe I'll be able to send you the
patches for approval next week.

Thank you,
Greg
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkvD5D8ACgkQ0sRouByUApC4AgCfS/QrY7PTkegsFRJiEE5WdxH5
GcIAoIdY7KkdUV+ORjj0G9iJcipt31jK
=HuD1
-----END PGP SIGNATURE-----
_______________________________________________
freebsd-ports(a)freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"

From: Tsurutani Naoki on
Greg Larkin <glarkin(a)freebsd.org> wrote:

> Can you run the following command for me on your system?
>
> ldconfig -r | grep gcc45

% ldconfig -r | grep gcc45
search directories:
/lib:/usr/lib:/usr/lib/compat:/usr/local/lib:/usr/local/lib/compat/pkg:/usr/site/lib:/usr/loc
al/lib/compat:/usr/local/lib/evolution/2.28:/usr/local/lib/gcc/i386-portbld-
freebsd8.0/3.4.6:/usr/local/lib/gcc44:/usr/local/lib/gcc45:/usr/local/lib/gegl-
0.0:/usr/local/lib/graphviz:/usr/local/lib/mysql:/usr/local/lib/nss:/usr/local/lib/pth:/usr/l
ocal/lib/qt4:/usr/local/lib/wine:/usr/local/lib/zsh
1159:-lstdc++.6 => /usr/local/lib/gcc45/libstdc++.so.6
1160:-lmudflap.0 => /usr/local/lib/gcc45/libmudflap.so.0
1161:-lmudflapth.0 => /usr/local/lib/gcc45/libmudflapth.so.0
1162:-lssp.0 => /usr/local/lib/gcc45/libssp.so.0
1163:-lgcc_s.1 => /usr/local/lib/gcc45/libgcc_s.so.1
1164:-lgfortran.3 => /usr/local/lib/gcc45/libgfortran.so.3
1165:-lffi.4 => /usr/local/lib/gcc45/libffi.so.4
1166:-lgcj.11 => /usr/local/lib/gcc45/libgcj.so.11
1167:-lgij.11 => /usr/local/lib/gcc45/libgij.so.11
1168:-lgcj-tools.11 => /usr/local/lib/gcc45/libgcj-tools.so.11
1169:-lgomp.1 => /usr/local/lib/gcc45/libgomp.so.1

% ldconfig -r | grep libstdc++.so.6
80:-lstdc++.6 => /usr/lib/libstdc++.so.6
848:-lstdc++.6 => /usr/local/lib/compat/pkg/libstdc++.so.6
1146:-lstdc++.6 => /usr/local/lib/gcc/i386-portbld-freebsd8.0/3.4.6/libstdc++.so.6
1148:-lstdc++.6 => /usr/local/lib/gcc44/libstdc++.so.6
1159:-lstdc++.6 => /usr/local/lib/gcc45/libstdc++.so.6

Setting LD_LIBRARY_PATH=/usr/local/lib/gcc45 before running pdftk
can avoid this error.
I am not familiar with rpath, but something is wrong arround it, I think.
gcc can embed some rpath infomation when building pdftk,
or we must add some linker option about rpath in building pdftk.
In any case, pdftk must have runtime dependence on gcc45...


---
$BDaC+D><y(B
turutani(a)scphys.kyoto-u.ac.jp
_______________________________________________
freebsd-ports(a)freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"

From: Greg Larkin on
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tsurutani Naoki wrote:
> Greg Larkin <glarkin(a)freebsd.org> wrote:
>
>> Can you run the following command for me on your system?
>>
>> ldconfig -r | grep gcc45
>
> % ldconfig -r | grep gcc45
> search directories:
[...]
>
> % ldconfig -r | grep libstdc++.so.6
> 80:-lstdc++.6 => /usr/lib/libstdc++.so.6
> 848:-lstdc++.6 => /usr/local/lib/compat/pkg/libstdc++.so.6
> 1146:-lstdc++.6 => /usr/local/lib/gcc/i386-portbld-freebsd8.0/3.4.6/libstdc++.so.6
> 1148:-lstdc++.6 => /usr/local/lib/gcc44/libstdc++.so.6
> 1159:-lstdc++.6 => /usr/local/lib/gcc45/libstdc++.so.6
>
> Setting LD_LIBRARY_PATH=/usr/local/lib/gcc45 before running pdftk
> can avoid this error.
> I am not familiar with rpath, but something is wrong arround it, I think.
> gcc can embed some rpath infomation when building pdftk,
> or we must add some linker option about rpath in building pdftk.
> In any case, pdftk must have runtime dependence on gcc45...
>
>
> ---
> $BDaC+D><y(B
> turutani(a)scphys.kyoto-u.ac.jp

Hi Tsurutani,

Ok, I see why there's a problem now. My linker hints were set up in
such a way that /usr/local/lib/gcc45 appeared before /usr/lib, so I
didn't have the libstdc++.so.6 problem. However, that's not a normal
configuration, so we have to fix this another way.

Gerald, you ran into this problem a while back and posted a question
about it here:
http://www.mail-archive.com/freebsd-ports(a)freebsd.org/msg23261.html

There are some other folks with the same problem here (old thread):
http://gcc.gnu.org/ml/gcc/2001-03/msg01069.html

Did you ever decide on a resolution for this problem? Should I simply
create a wrapper script for pdftk that sets LD_LIBRARY_PATH, or is the
rpath solution better?

Tsurutani, in the mean time, can you run pdftk with LD_LIBRARY_PATH set
until we decide on the solution?

Thank you,
Greg
- --
Greg Larkin

http://www.FreeBSD.org/ - The Power To Serve
http://www.sourcehosting.net/ - Ready. Set. Code.
http://twitter.com/sourcehosting/ - Follow me, follow you
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iD8DBQFLxLRM0sRouByUApARAvfDAJ93Bm4wyhUlpl7PTZR6Ulpg9J31SQCfTAIu
9QRjwZAlBepsnFpWuGuMvZ8=
=KPDi
-----END PGP SIGNATURE-----

_______________________________________________
freebsd-ports(a)freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"

From: Greg Larkin on
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tsurutani Naoki wrote:
[...]
> 2. FreeBSD/amd64 8-STABLE, with LANG=ja_JP.eucJP : building failure.
> gmake -C "/usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs/com/lowagie/text";
> gmake[2]: Entering directory `/usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs/com/lowagie/text'
> /usr/local/bin/gcj45 -L/usr/local/lib -O2 -w --encoding=UTF-8 --classpath="/usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs" -c Anchor.java -o Anchor.o
> /usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs/com/lowagie/text/markup/MarkupParser.java:186: error: String literal is not properly closed by a double-quote
> if (value.startsWith("$B")(B)
> ^^^
> ...
> 112 problems (112 errors)
> gmake[2]: *** [Anchor.o] Error 1
> gmake[2]: Leaving directory `/usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs/com/lowagie/text'
> gmake[1]: *** [itext] Error 2
> gmake[1]: Leaving directory `/usr/local/work/usr/ports/print/pdftk/work/pdftk-1.41/java_libs'
> gmake: *** [java_libs] Error 2
> *** Error code 1
> (some messages are translated).

Hi Tsurutani,

As far as I can tell, this error is caused by the LANG setting which
changes the line:

if (value.startsWith("\""))

into something that results an unterminated string. I suppose the
backslash loses its meaning as an escape character.

I found some references to the same error in Google, posted by folks
trying to compile software with different LANG settings. I haven't
found a general fix yet, but maybe it can be done by building a
character string from individual, unescaped characters, then converting
them to a java.lang.String object for use in the .startsWith() function.

If you could submit a PR to support building with LANG=ja_JP.eucJP, I
would be very appreciative!

Thank you,
Greg
- --
Greg Larkin

http://www.FreeBSD.org/ - The Power To Serve
http://www.sourcehosting.net/ - Ready. Set. Code.
http://twitter.com/sourcehosting/ - Follow me, follow you
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iD8DBQFLxLcS0sRouByUApARAm4aAKC2xbzaY34Q1GAPCv+mVifG7raC0QCgtjwZ
eGEmwcVfVM5N2AbrfjDFJLs=
=cfvI
-----END PGP SIGNATURE-----

_______________________________________________
freebsd-ports(a)freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"

From: Gerald Pfeifer on
On Tue, 13 Apr 2010, Greg Larkin wrote:
> Ok, I see why there's a problem now. My linker hints were set up in
> such a way that /usr/local/lib/gcc45 appeared before /usr/lib, so I
> didn't have the libstdc++.so.6 problem. However, that's not a normal
> configuration, so we have to fix this another way.
>
> Gerald, you ran into this problem a while back and posted a question
> about it here:
> http://www.mail-archive.com/freebsd-ports(a)freebsd.org/msg23261.html
>
> There are some other folks with the same problem here (old thread):
> http://gcc.gnu.org/ml/gcc/2001-03/msg01069.html
>
> Did you ever decide on a resolution for this problem? Should I simply
> create a wrapper script for pdftk that sets LD_LIBRARY_PATH, or is the
> rpath solution better?

The way I have addressed this for C, C++ and from what I can tell all
relevant Fortran users of USE_GCC= is by setting the following

CFLAGS+= -Wl,-rpath=${LOCALBASE}/lib/${_GCC_BUILD_DEPENDS}
LDFLAGS+= -Wl,-rpath=${LOCALBASE}/lib/${_GCC_BUILD_DEPENDS}

in Mk/bsd.gcc.mk.

Apparently pdftk does not use LDFLAGS (and it not using CFLAGS for Java
code is somewhat expected). Is there some similar variable we could set
for Java code? JFLAG, JAVAFLAGS,...? Or could you inject the use of
CFLAGS or LDFLAGS during linking?

Gerald
_______________________________________________
freebsd-ports(a)freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"