Re: Backend issues on NetBSD

[Fred Kiefer]

> Am 18.02.2018 um 19:19 schrieb Riccardo Mottola <address@hidden>:
> 
> I just rebuilt gnustep-back on NetBSD 6.1.5 x86 and clang where it builds, 
> but then fails tu load/run with:
> 
> 2018-02-18 17:48:19.355 Ink[22510:3099672676] NSApplication.m:317  Assertion 
> failed in BOOL initialize_gnustep_backend().  Can't load object file from 
> backend at path /System/Library/Bundles/libgnustep-back-026.bundle
> Ink: Uncaught exception NSInternalInconsistencyException, reason: 
> NSApplication.m:317  Assertion failed in BOOL initialize_gnustep_backend().  
> Can't load object file from backend at path 
> /System/Library/Bundles/libgnustep-back-026.bundle

I take it that this is with the cairo backend, as later you mention the xlib 
backend? Could you please be more clear about this as you are often using 
unusual setups and it is hard to tell which of these it is this time.

> The last time i checked this setup, it worked. It is a slightly older NetBSd 
> than I generally use but it uses clang instead of gcc
> 
> Relevant sections in the configure log seem also fine to me, pkgconfig gets 
> detected and the config output is below.
> Cairo is detected, not freetype, so it gets switched to xlib, but still I 
> guess it should work
> 
> checking for XcursorImageCreate in -lXcursor... yes
> 
> Xcursor is explicitely found!

Why do you think the issue is Xcursor related? Was there more output that 
hinted in this direction?
The most interesting information here would be the result of an ldd on 
/System/Library/Bundles/libgnustep-back-026.bundle. That way we would know 
which libraries are linked into the backend.


> An issue might be:
> configure:3954: checking whether -R must be followed by a space
> configure:3970: clang -o conftest -g -O2  -I/System/Library/Headers 
> -I/usr/pkg/include  -L/System/Library/Libraries/gnu-gnu-gnu 
> -L/System/Library/Libraries -Wl,-R/usr/pkg/lib -L/usr/pkg/lib conftest.c  
> -R/usr/X11R7/lib >&5
> clang: warning: argument unused during compilation: '-R/usr/X11R7/lib'
> 
> what would be this -R/usr/X11R7/lib ? Shouldn't it always be 
> -Wl,-R/usr/X11R7/lib ??

Could be that this check is wrong, but most likely it was never needed for 
clang anyway.


> The xlib backend should build just fine.
> I got around the linking issue by adding LDFLAGS="-Wl,-R/usr/X11R7/lib 
> -L/usr/X11R7/lib" however then no Fonts gets displayed! It may be the same 
> issue that Cairo has: wrong library.

Now you are talking about a backend that starts up, but doesn’t display fonts. 
But are the following lines from the configure output for cairo or xlib? The 
actual output should be similar in both cases.
And the lib backend should fall back to some other font system if freetype 
isn’t available. There should always be a way to display strings for this 
backend. On the other hand the cairo backend should not build at all without 
freetype and fall back to xlib. I am a bit confused what happened here.

> Further question is: why doesn't the cairo backend work out? This is the 
> failing test:
> 
> configure:5300: checking ft2build.h usability
> configure:5300: clang -c -g -O2  -I/usr/X11R7/include  
> -I/System/Library/Headers
> -I/usr/pkg/include -I/usr/pkg/include/freetype2  conftest.c >&5
> In file included from conftest.c:62:
> /usr/X11R7/include/ft2build.h:34:10: fatal error: 
> 'freetype/config/ftheader.h' f
> ile not found
> #include <freetype/config/ftheader.h>
> 
> I have both a freetype package as a system freetype I guess the "order" is 
> frong, since the system path is before pkg, it gets detected. pkg-config is 
> returning the flags for the pkg one and not the system one, but then the 
> tests finds the system one and fails. I think somewhere we append these flags 
> instead of prepending them!

Why do you have this library installed twice? And do you have the headers also 
twice? Then this test shouldn’t fail. It rather looks like there aren’t any 
header files for freetype at all. Could you please check where these files are?
I think I understand now what is going on. You have a half installed freetype 
that comes with your X installation and this gets found before the properly 
installed one. The change you made below won’t help here. You rather should 
clean up your installation. Prepending the freetype paths, as you did in your 
final change, might be a valid workaround. This could be a change we should 
integrate into back.


> I found this in our configure.ac
> 
>  GRAPHIC_CFLAGS="$X_CFLAGS $GRAPHIC_CFLAGS"
>  CPPFLAGS="$X_CFLAGS $CPPFLAGS"
>  # Note: Don't include $X_LIBS in LDFLAGS as it may conflict with
>  # other libraries installed in /sw, /sw/lib/freetyp2/lib
>  #GRAPHIC_LFLAGS="$X_LIBS $GRAPHIC_LFLAGS"
>  # But do include it here just to find the extra x libraries
>  LDFLAGS="$X_LIBS $LDFLAGS"
>  LIBS="-lX11 $X_EXTRA_LIBS $LIBS"
> 
> First, note that there is a comment... and then we exactly to what the 
> comment says not to do! Since freetype is excactly the library I am having an 
> issue with....
> Both LDFLAGS and CPPFLAGS are appended and not prepended, I did change that.

Yes, the comment conflicts with the actual code. But are you really sure that 
this was your actual problem? Or would the change you mention below be 
sufficient?
And actually the X paths are prepended here and this is also what you are doing 
below the the freetype paths.

> Then I think also this needs to be fixed:
>    CPPFLAGS="${FREETYPE_CFLAGS} ${CPPFLAGS}“

Yes, this could be something that is needed to get the configure script working 
on broken systems.

> This finally worked for cairo (did not recheck if xlib works though). I 
> acclude the configure patch. I did not commit it yet and am open for 
> discussion, might this cause unforseen other issues?
> What do you guys thik?
> And the note about X_LIBS that somebody added?


> 
> narsil: {9} ./configure
> checking build system type... i386-unknown-netbsdelf6.1
> checking host system type... i386-unknown-netbsdelf6.1
> checking target system type... i386-unknown-netbsdelf6.1
> checking for gcc... clang
> 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... no
> checking for suffix of object files... o
> checking whether we are using the GNU C compiler... yes
> checking whether clang accepts -g... yes
> checking for clang option to accept ISO C89... none needed
> checking how to run the C preprocessor... clang -E
> checking for grep that handles long lines and -e... /usr/bin/grep
> checking for egrep... /usr/bin/grep -E
> checking for ANSI C header files... yes
> checking for sys/types.h... yes
> checking for sys/stat.h... yes
> checking for stdlib.h... yes
> checking for string.h... yes
> checking for memory.h... yes
> checking for strings.h... yes
> checking for inttypes.h... yes
> checking for stdint.h... yes
> checking for unistd.h... yes
> checking syslog.h usability... yes
> checking syslog.h presence... yes
> checking for syslog.h... yes
> checking for syslog... yes
> checking for X... libraries /usr/X11R7/lib, headers /usr/X11R7/include
> checking whether -R must be followed by a space... neither works
> checking for gethostbyname... yes
> checking for connect... yes
> checking for remove... yes
> checking for shmat... yes
> checking for IceConnectionNumber in -lICE... yes
> checking for pkg-config... /usr/pkg/bin/pkg-config
> checking pkg-config is at least version 0.9.0... yes
> checking for main in -lXext... yes
> checking for main in -lXt... yes
> checking for main in -lXmu... yes
> checking for XFixesSelectSelectionInput in -lXfixes... yes
> checking for XcursorImageCreate in -lXcursor... yes
> checking for X11/extensions/shape.h... yes
> checking for XShapeCombineMask in -lXext... yes
> checking for X11/extensions/sync.h... yes
> checking for X11 function prototypes... yes
> checking DPS/dpsclient.h usability... no
> checking DPS/dpsclient.h presence... no
> checking for DPS/dpsclient.h... no
> checking DPS/dpsNXargs.h usability... no
> checking DPS/dpsNXargs.h presence... no
> checking for DPS/dpsNXargs.h... no
> checking for freetype2... yes
> checking ft2build.h usability... no
> checking ft2build.h presence... no
> checking for ft2build.h... no
> checking for xft... yes
> checking for XftDrawStringUtf8 in -lXft... yes
> checking for XftPatternGetString in -lXft... no
> checking for Xutf8LookupString... yes
> checking for FcPatternCreate in -lfontconfig... yes
> checking fontconfig/fontconfig.h usability... yes
> checking fontconfig/fontconfig.h presence... yes
> checking for fontconfig/fontconfig.h... yes
> checking for glXMakeContextCurrent in -lGL... yes
> checking GL/glx.h usability... yes
> checking GL/glx.h presence... yes
> checking for GL/glx.h... yes
> checking for GLX_RGBA_TYPE... yes
> checking for usleep... yes
> checking for X11/extensions/XShm.h... yes
> checking for shmctl... yes
> checking for XInternAtoms in -lX11... yes
> checking for main in -lgdi32... no
> checking for main in -lmsimg32... no
> checking for main in -limm32... no
> checking for main in -lopengl32... no
> checking windows.h usability... no
> checking windows.h presence... no
> checking for windows.h... no
> checking for libart2... none
> checking for main in -lart_lgpl_2... no
> checking for cairo... yes
> checking for cairo-ft... yes
> checking for cairo-xlib... yes
> checking for cairo-win32... no
> checking for cairo-glitz... no
> checking for fontconfig... yes
> checking for cairo_win32_surface_create in -lcairo... no
> checking for XRenderFindVisualFormat in -lXrender... yes
> checking Backend Server... x11
> checking Backend Graphics... cairo
> configure: WARNING: can't find freetype, required for graphics=cairo
> configure: Switching to xlib
> checking whether the compiler supports -Wdeclaration-after-statement... yes
> checking Backend name... back
> configure: creating ./config.status
> config.status: creating back.make
> config.status: creating config.make
> config.status: creating config.h
> config.status: config.h is unchanged
> 
> <flags-configure.patch>_______________________________________________
> Gnustep-dev mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/gnustep-dev