Re: [ft-devel] a couple of warnings from 2.5.4 with mingw/darwinx

[suzuki toshiya]

Hi,

> Actually that's what I have. Sorry about the delay. My build was driven
> by the crosscompile framework.
>
> http://fedoraproject.org/wiki/MinGW/CrossCompilerFramework#Running_the_..2Fconfigure_script

> It is a bit complicated, but I think in a nutshell it means running
> configure with CC="darwinx-gcc -arch ppc -arch i386 -arch x86_64"
> i.e. building "fat" object files which are triarch.

I ask you for more detailed information to reproduce the problem,
I could not reproduce the problem. However, it seems that the cross
toolchain linked from the web page in above could not make configure
script to recognize the linker appropriately. Even if the configuration
finishes successfully, "make" will aborted as "ld: unknown option: -soname",
like this.

libtool: link: darwinx-gcc -arch ppc -arch i386 -arch x86_64 -shared  
-fno-common -DPIC  \
        ... /psnames.o   -lc  -arch ppc -arch i386 -arch x86_64 \
        -Wl,-soname -Wl,libfreetype.6.dylib \
        -o 
/home/sssa/redhat/BUILD/freetype2-current/macosx/freetype-2.5.5/objs/.libs/libfreetype.6.dylib
collect2: ld returned 1 exit status
ld: unknown option: -soname
collect2: ld returned 1 exit status
ld: unknown option: -soname
collect2: ld returned 1 exit status
darwinx-lipo: can't open input file: /tmp/ccaCwxLC.out (No such file or 
directory)
config.mk:55: recipe for target 
'/home/sssa/redhat/BUILD/freetype2-current/macosx/freetype-2.5.5/objs/libfreetype.la'
 failed


Anyway, below are my testings about the configuration process.

Case 1) using CC="..." only
----------------------------

Configuration itself does not finish, because the system could not
recognize that we are cross building.

$ env CC="darwinx-gcc -arch ppc -arch i386 -arch x86_64" ./configure

FreeType build system -- automatic system detection

The following settings are used:

  platform                    unix
  compiler                    darwinx-gcc -arch ppc -arch i386 -arch x86_64
  configuration directory     ./builds/unix
  configuration rules         ./builds/unix/unix.mk

If this does not correspond to your system or settings please remove the file
`config.mk' from this directory then read the INSTALL file for help.

Otherwise, simply type `make' again to build the library,
or `make refdoc' to build the API reference (this needs python >= 2.6).

cd builds/unix; /bin/sh ./configure
checking build system type... x86_64-unknown-linux-gnu
checking host system type... x86_64-unknown-linux-gnu
checking for gcc... darwinx-gcc -arch ppc -arch i386 -arch x86_64
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... configure: error: in 
`/home/sssa/redhat/BUILD/freetype2-current/freetype2/builds/unix':
configure: error: cannot run C compiled programs.
If you meant to cross compile, use `--host'.
See `config.log' for more details
builds/unix/detect.mk:86: recipe for target 'setup' failed
make: *** [setup] Error 1

Case 2) using CC="..." and "--build=x86_64-linux-gnu"
--------------------------------------------------------

Configuration itself does not finish, because the system could not
recognize that we are cross building.

$ env CC="darwinx-gcc -arch ppc -arch i386 -arch x86_64" ./configure 
--build=x86_64-linux-gnu

FreeType build system -- automatic system detection

The following settings are used:

  platform                    unix
  compiler                    darwinx-gcc -arch ppc -arch i386 -arch x86_64
  configuration directory     ./builds/unix
  configuration rules         ./builds/unix/unix.mk

If this does not correspond to your system or settings please remove the file
`config.mk' from this directory then read the INSTALL file for help.

Otherwise, simply type `make' again to build the library,
or `make refdoc' to build the API reference (this needs python >= 2.6).

cd builds/unix; /bin/sh ./configure  '--build=x86_64-linux-gnu'
checking build system type... x86_64-pc-linux-gnu
checking host system type... x86_64-pc-linux-gnu
checking for gcc... darwinx-gcc -arch ppc -arch i386 -arch x86_64
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... configure: error: in 
`/home/sssa/redhat/BUILD/freetype2-current/freetype2/builds/unix':
configure: error: cannot run C compiled programs.
If you meant to cross compile, use `--host'.
See `config.log' for more details
builds/unix/detect.mk:86: recipe for target 'setup' failed
make: *** [setup] Error 1

Case 3) using CC="..." and "--host=i686-apple-darwin10"
--------------------------------------------------------

No error in "checking whether cpp computation of bit length in ftconfig.in 
works".

$ env CC="darwinx-gcc -arch ppc -arch i386 -arch x86_64" ./configure 
--host=i686-apple-darwin10

FreeType build system -- automatic system detection

The following settings are used:

  platform                    unix
  compiler                    darwinx-gcc -arch ppc -arch i386 -arch x86_64
  configuration directory     ./builds/unix
  configuration rules         ./builds/unix/unix.mk

If this does not correspond to your system or settings please remove the file
`config.mk' from this directory then read the INSTALL file for help.

Otherwise, simply type `make' again to build the library,
or `make refdoc' to build the API reference (this needs python >= 2.6).

cd builds/unix; /bin/sh ./configure  '--host=i686-apple-darwin10'
checking build system type... x86_64-unknown-linux-gnu
checking host system type... i686-apple-darwin10
checking for i686-apple-darwin10-gcc... darwinx-gcc -arch ppc -arch i386 -arch 
x86_64
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... yes

        <snip>

checking for unistd.h... (cached) yes
checking for an ANSI C-conforming const... yes
checking size of int... 4
checking size of long... 8
checking whether cpp computation of bit length in ftconfig.in works... yes
checking for stdlib.h... (cached) yes
checking for unistd.h... (cached) yes
checking for sys/param.h... yes

        <snip>

Regards,
mpsuzuki

P.S.
<whine>
I feel awkward saying this, I think the website is poorly documented
about the cross building. Even I could not find the note about how
to use GNU configure's options "--host", "--build" and "--target".
Maybe it would be sufficient to guide Fedora users to work with MinGW
toolchain on Fedora, but I feel using this document for Mac OS X cross
development (or other generic context) is something like an attempt to
use the memo telling "how to catch a taxi" for a hitchhike. When I
wrote INSTALL.CROSS in 8 years ago, already there were so many
precarious web pages teaching people as "CC=blahblahblah-gcc" is good
for cross building. It is miserable to see that still the precarious
web pages about the cross building are actively recreated.
</whine>



Hin-Tak Leung wrote:
> Actually that's what I have. Sorry about the delay. My build was driven
> by the crosscompile framework. 
> 
> http://fedoraproject.org/wiki/MinGW/CrossCompilerFramework#Running_the_..2Fconfigure_script
> 
> It is a bit complicated, but I think in a nutshell it means running
> configure with CC="darwinx-gcc -arch ppc -arch i386 -arch x86_64"
> i.e. building "fat" object files which are triarch.
> 
> for your reference, here is the freetype headers shipped by xquartz.
> I have also what apple did with freetype (they shipped it "officially" for 
> 10.6/10.7,
> I think the xquartz people are essentially also apple's) e.g.
> https://github.com/phracker/MacOSX-SDKs/blob/master/MacOSX10.7.sdk/usr/X11/include/freetype2/freetype/config/ftconfig.h
> 
> xquartz's freetype headers seems to have extra assembly codes;
> whereas the apple SDKs' freetype headers seems to have FT_MACINTOSH always
> enabled. Am I correct?
> 
> --------------------------------------------
> On Thu, 29/1/15, suzuki toshiya <address@hidden> wrote:
>  
>  I tried with Fedora's
>  cross compiler (for darwinx),
>  and could not
>  reproduce the error. Which toolchain
>  I
>  should try in next?
>  
>  Regards,
>  mpsuzuki
>  
>  suzuki toshiya wrote:
>  > Hi,
>  > 
>  > Yet I could not reproduce the problem on
>  my Mac OS X,
>  > I think I should try on
>  the cross building system.
>  > However,
>  I'm unfamiliar with the defacto standard of
>  > the cross compiler for DarwinX on other
>  platform.
>  > 
>  >
>  Considering that you also mentioned "for mingw", I
>  guess
>  > you're working on some unix
>  like platform. Your cross
>  > compiler for
>  DarwinX is built by yourself? Or you
>  >
>  installed some prebuilt packages? The prebuilt package
>  > I could find was only for Fedora:
>  > 
> http://build1.vanpienbroek.nl/fedora-cross-darwinx/fedora-cross-darwinx.repo
>  > 
>  > Please let me know
>  more detail about your platform working for DarwinX.
>  > 
>  > Regards,
>  > mpsuzuki
>  > Hin-Tak
>  Leung wrote:
>  >> I have got to the
>  bottom of the latter issue; I mentioned that I was
>  cross-compiling,
>  >> for apple, right?
>  It really needs the --enable-biarch-config switch:
>  >>
>  >> without
>  --enable-biarch-config:
>  >>
>  >> checking size of long... 8
>  >> checking whether cpp computation of
>  bit length in ftconfig.in works... no
>  >>
>  >> with
>  --enable-biarch-config:
>  >>
>  >> checking size of long... 8
>  >> checking whether cpp computation of
>  bit length in ftconfig.in works... broken but use it
>  >>
>  >> I don't
>  know where it gets the size of long from. The apple
>  compiler
>  >> front end is quite
>  interesting in that it drives the individual
>  architectures
>  >> and put all three (or
>  4) outputs from 32-bit powerpc, 32-bit intel, 64-bit intel
>  into 1 "object" files.
>  >>
>  And the warnings come from the "under the hood"
>  compile of the 32-bit outputs.
>  >>
>  >> The 32-bit powerpc and 32-bit intel
>  arch have size of long = 4. If I use
>  >> the individual single architecture
>  front end, then it shows "4" and "yes"
>  respectively.
>  >>
>  >> I am wondering whether
>  --enable-biarch-config should be better documented,
>  >> and/or made the default for Mac OS X?
>  (though it probably make less sense now since
>  >> apple has moved to 64-bit intel
>  somewhat exclusively lately)
>  >>
>  >> Also, can it not use stdint.h and
>  int64_t directly? "long" is rather vague :-).
>  >>
>  >>
>  --------------------------------------------
>  >> On Sat, 17/1/15, Werner LEMBERG <address@hidden> wrote:
>  >>  
>  >>  Hello
>  >>  Hin-Tak!
>  >> 
>  
>  >>  
>  >> 
>  >
>  >> 
>  /usr/i686-w64-mingw32/sys-root/mingw/include/harfbuzz/hb-common.h:309:26:
>  >>  > warning: ISO C restricts
>  enumerator values
>  >>  to range of
>  'int'
>  >>  >
>  >>  [-Wpedantic]
>  >>  >   
>  >>  _HB_SCRIPT_MAX_VALUE    =
>  HB_TAG_MAX, /*< skip
>  >>  >*/
>  >>  >                 
>     
>  >>   
>     ^
>  >>  
>  >>  This is
>  >>  a
>  known but harmless issue, and work-arounds seem to be
>  >>  very
>  >> 
>  inelegant, IIRC.
>  >>  
>  >>  >
>  >> 
>  /root/rpmbuild/BUILD/freetype-2.5.4/src/base/fttrigon.c:74:
>  >>  warning:
>  >> 
>  > right shift count >= width
>  >>  of type
>  >> 
>  
>  >>  This looks
>  >>  strange.  Here's the
>  corresponding code line:
>  >>  
>  >>    val = (FT_Fixed)( (
>  >>  (FT_Int64)val * FT_TRIG_SCALE +
>  0x40000000UL ) >> 32
>  >>  );
>  >>  
>  >>  We
>  explicitly convert to
>  >>  a 64bit
>  entity (`FT_Int64'), and this should
>  >>  allow a shifting by 32 bits...  So
>  the basic
>  >>  question is whether
>  >>  `FT_Int64' is
>  >>  *indeed* a 64bit type.
>  >>  
>  >>  
>  >>      Werner
>  >>  
>  >>
>  >>
>  _______________________________________________
>  >> Freetype-devel mailing list
>  >> address@hidden
>  >> https://lists.nongnu.org/mailman/listinfo/freetype-devel
>  >
>  
>  > 
>  >
>  _______________________________________________
>  > Freetype-devel mailing list
>  > address@hidden
>  > https://lists.nongnu.org/mailman/listinfo/freetype-devel