Compiling trunk on Windows

[Yoann Roman-2]

I'm trying to get Fribidi compiled on Windows for use with Python via
Pyfribidi (http://pyfribidi.sourceforge.net). I was able to compile
the January 15th 0.19.1 archive with VS2003 & VS2008 by using Christophe
Gisquet's files/notes (http://tinyurl.com/cvzfac). I also got that
version to compile with Cygwin using configure / make.

Unfortunately, when I pass that version some Arabic text, I'm getting
back invalid UTF8 sequences (2-bytes for a 3-byte codepoint). I didn't
have that problem with the 0.10.9 release also compiled with VS2003, but
that version didn't do the Arabic joining.

All that to say I'm trying to see if the fribidi2 trunk code works
better. Problem is I can't get it to compile on Windows. I need to get
the .tab.i files generated, but I can't get anything under /gen.tab to
compile correctly in VS2003, VS2005, or VS2008. I tried the Cygwin
approach, but autoconf barfs at configure.ac:

configure.ac:45: error: possibly undefined macro: AM_INIT_AUTOMAKE
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.
configure.ac:78: error: possibly undefined macro: AC_LIBTOOL_WIN32_DLL
configure.ac:79: error: possibly undefined macro: AC_PROG_LIBTOOL
configure.ac:86: error: possibly undefined macro: AM_CONDITIONAL

Any suggestions?

Thanks,

--
Yoann Roman

_______________________________________________
fribidi mailing list
fribidi@...
http://lists.freedesktop.org/mailman/listinfo/fribidi

[Yoann Roman-2]

> Being in a similar spot I succeeded recently in building fribidi2 from
> CVS using the mingw32 cross compiler on Linux. To call the DLL from
> python I used the ctypes module.

Thanks for the rapid response. After some more playing around with
compiler flags and Windows fixes, I got the trunk code to compile the
generators, run them, and compile fribidi.dll with VS2003. I can't get
ctypes to work with that DLL, but Pyfribidi seems to load fine.

My problem now is that it's still returning an invalid UTF-8 sequence
for the Arabic text I'm giving it, but that's for another message.

Thanks,

--
Yoann Roman

_______________________________________________
fribidi mailing list
fribidi@...
http://lists.freedesktop.org/mailman/listinfo/fribidi

[Behdad Esfahbod-3]

Hi Yoann,

The fribidi2 trunk is not much different from the release 0.19.1 version, so I
suggest you focus on figuring out why you are getting the wrong output from
0.19.1.

Regards,

behdad

Yoann Roman wrote:

> I'm trying to get Fribidi compiled on Windows for use with Python via
> Pyfribidi (http://pyfribidi.sourceforge.net). I was able to compile
> the January 15th 0.19.1 archive with VS2003 & VS2008 by using Christophe
> Gisquet's files/notes (http://tinyurl.com/cvzfac). I also got that
> version to compile with Cygwin using configure / make.
>
> Unfortunately, when I pass that version some Arabic text, I'm getting
> back invalid UTF8 sequences (2-bytes for a 3-byte codepoint). I didn't
> have that problem with the 0.10.9 release also compiled with VS2003, but
> that version didn't do the Arabic joining.
>
> All that to say I'm trying to see if the fribidi2 trunk code works
> better. Problem is I can't get it to compile on Windows. I need to get
> the .tab.i files generated, but I can't get anything under /gen.tab to
> compile correctly in VS2003, VS2005, or VS2008. I tried the Cygwin
> approach, but autoconf barfs at configure.ac:
>
> configure.ac:45: error: possibly undefined macro: AM_INIT_AUTOMAKE
>       If this token and others are legitimate, please use m4_pattern_allow.
>       See the Autoconf documentation.
> configure.ac:78: error: possibly undefined macro: AC_LIBTOOL_WIN32_DLL
> configure.ac:79: error: possibly undefined macro: AC_PROG_LIBTOOL
> configure.ac:86: error: possibly undefined macro: AM_CONDITIONAL
>
> Any suggestions?
>
> Thanks,
>

_______________________________________________
fribidi mailing list
fribidi@...
http://lists.freedesktop.org/mailman/listinfo/fribidi

[Yoann Roman-2]

> Hi Yoann,
>
> The fribidi2 trunk is not much different from the release 0.19.1
> version, so I suggest you focus on figuring out why you are getting
> the wrong output from 0.19.1.

Any tips on where in the fribidi code I should start looking?

Thanks,

--
Yoann Roman

_______________________________________________
fribidi mailing list
fribidi@...
http://lists.freedesktop.org/mailman/listinfo/fribidi