Re: [Gtk-osx-devs] Latest stable libxml2 fails build on OS X 10.4

[John Ralls <jralls ceridwen us>]

On Jul 19, 2010, at 12:35 PM, Richard Procter wrote:

> Hi John,
> 
>> Well, that's annoying. I had to upgrade it because the latest  
>> SnowLeopard upgrade didn't like libxml2-2.6. I guess I'll have to  
>> add a branches["libxml2"] entry in jhbuildrc-custom to use the  
>> older version. (Good thing I took out the checksum, I guess. At  
>> least that won't choke, though something else will, I'm sure.)
> 
> Dependencies are hell - I'm glad I don't have to deal with them -  
> thanks again for the job you're doing.
> 
>> I'm not where I can use my 10.4 build machine. Could you add
>> branches["libxml2"]="libxml2/2.6/libxml2-2.6.30.tar.bz2"
>> to your jhbuildrc-custom and see if that works? If it does (and you  
>> report back before I get home in a couple of weeks), I'll commit a  
>> change to jhbuildrc so that it's automatic on Tiger.
>> 
>> Have you tried building gtk-2.90 on Tiger?
>> 
> 
> 
> Tiger is not a development target of mine, so this is not a big issue  
> for me at least (I just happen to have a Tiger machine).
> 
> I've tried building 2.20 and found the the following issues (if you'd  
> specifically like me to have a crack at 2.90 I can give it go this  
> weekend):
> 
> 1) the libtiff repository is unresponsive.
> 
> *** Checking out libtiff *** [8/17]
> curl --continue-at - -L ftp://ftp.remotesensing.org/pub/libtiff/ 
> tiff-3.9.1.tar.gz -o /Volumes/Vault/tmp/gtk/source/pkgs/ 
> tiff-3.9.1.tar.gz
> 
> curl: (56) FTP response reading failed
> 
> 2) OS X 10.4 barfs on NSUInteger:
> 
> libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I.. -DG_LOG_DOMAIN=\"Gtk 
> \" -DGTK_LIBDIR=\"/Volumes/Vault/tmp/gtk/inst/lib\" -DGTK_DATADIR=\"/ 
> Volumes/Vault/tmp/gtk/inst/share\" -DGTK_DATA_PREFIX=\"/Volumes/Vault/ 
> tmp/gtk/inst\" -DGTK_SYSCONFDIR=\"/Volumes/Vault/tmp/gtk/inst/etc\" - 
> DGTK_VERSION=\"2.20.1\" -DGTK_BINARY_VERSION=\"2.10.0\" -DGTK_HOST= 
> \"powerpc-apple-darwin8.11.0\" -DGTK_COMPILATION -DGTK_PRINT_BACKENDS= 
> \"file,cups\" "-DGTK_PRINT_PREVIEW_COMMAND=\"open -a /Applications/ 
> Preview.app %f\"" -I.. -I../gtk -I.. -I../gdk -I../gdk -I../gdk- 
> pixbuf -I../gdk-pixbuf -DGDK_DISABLE_DEPRECATED - 
> DGTK_DISABLE_DEPRECATED -DGTK_FILE_SYSTEM_ENABLE_UNSUPPORTED - 
> DGTK_PRINT_BACKEND_ENABLE_UNSUPPORTED -DG_DISABLE_CAST_CHECKS - 
> D_REENTRANT -I/Volumes/Vault/tmp/gtk/inst/include/glib-2.0 -I/Volumes/ 
> Vault/tmp/gtk/inst/lib/glib-2.0/include -I/Volumes/Vault/tmp/gtk/inst/ 
> include/pango-1.0 -I/Volumes/Vault/tmp/gtk/inst/include/cairo -I/ 
> Volumes/Vault/tmp/gtk/inst/include/pixman-1 -I/Volumes/Vault/tmp/gtk/ 
> inst/include/freetype2 -I/Volumes/Vault/tmp/gtk/inst/include/libpng12  
> -I/Volumes/Vault/tmp/gtk/inst/include/atk-1.0 -xobjective-c -I/ 
> Volumes/Vault/tmp/gtk/inst/include -DG_DISABLE_SINGLE_INCLUDES - 
> DATK_DISABLE_SINGLE_INCLUDES -DGDK_PIXBUF_DISABLE_SINGLE_INCLUDES - 
> DGTK_DISABLE_SINGLE_INCLUDES -DGDK_PIXBUF_DISABLE_DEPRECATED -arch  
> ppc -mmacosx-version-min=10.4 -Wall -MT gtkquartz.lo -MD -MP - 
> MF .deps/gtkquartz.Tpo -c gtkquartz.c  -fno-common -DPIC -o .libs/ 
> gtkquartz.o
> gtkquartz.c: In function  
> '_gtk_quartz_set_selection_data_for_pasteboard':
> gtkquartz.c:263: error: 'NSUInteger' undeclared (first use in this  
> function)
> gtkquartz.c:263: error: (Each undeclared identifier is reported only  
> once
> gtkquartz.c:263: error: for each function it appears in.)
> gtkquartz.c:263: error: parse error before 'length'
> gtkquartz.c:269: error: 'length' undeclared (first use in this function)
> make[4]: *** [gtkquartz.lo] Error 1
> make[3]: *** [all-recursive] Error 1
> make[2]: *** [all] Error 2
> make[1]: *** [all-recursive] Error 1
> make: *** [all] Error 2
> *** Error during phase build of gtk+: ########## Error running make    
> *** [9/11]

Richard,

Thanks. What about the libxml2 fix? Does that work?

ftp.remotesensing.org does seem to be down. Their mirror at http://dl.maptools.org/dl/libtiff/ is up. 

Dang. There should be a define for NSUInteger. I'll have to write a patch for it. (You can fix it locally with 
#define unsigned int NSUInteger if you like.)

Regards,
John Ralls