Re: [Fink-users] xft2-dev-2.2.0-3 build fails due to incorrect include path in source code

[Stefan B. <st...@br...>]

Hello,

I see.  Thank you very much for the clarification.

Cheers,
Stefan

At 14:11 -0500 on 2015-1-25 Daniel Macks wrote:
 >
 > On Sun, 25 Jan 2015 13:03:51 -0500, Stefan Bruda <st...@br...> wrote:
 >   >
 >   > At 10:27 -0800 on 2015-1-24 Alexander Hansen wrote:
 >   > >
 >   > > > On Jan 24, 2015, at 9:37 AM, Brian Hayden <bha...@al...> wrote:
 >   > > >
 >   > > > When running:
 >   > > > fink install xft2-shlibs
 >   > > >
 >   > > > it crashes with:
 >   > > > libtool: compile: gcc -DHAVE_CONFIG_H -I. -I..
 > -I../include/X11/Xft -I../include/X11/Xft -I/sw/include
 > -I/sw/include/freetype2 -I/opt/X11/include -I/sw/include -Wall
 > -Wpointer-arith -Wstrict-prototypes -Wmissing-prototypes
 > -Wmissing-declarations -Wnested-externs -fno-strict-aliasing
 > -Wbad-function-cast -Wformat=2 -Wold-style-definition
 > -Wdeclaration-after-statement -g -O2 -MT xftglyphs.lo -MD -MP -MF
 > .deps/xftglyphs.Tpo -c xftglyphs.c -fno-common -DPIC -o
 > .libs/xftglyphs.o
 >   > > > xftglyphs.c:24:10: fatal error: 'freetype/ftoutln.h' file not found
 >   > > > #include <freetype/ftoutln.h>
 >   > > > ^
 >   > > > 1 error generated.
 >   > > > make[2]: *** [xftglyphs.lo] Error 1
 >   > > > make[1]: *** [all-recursive] Error 1
 >   > > > make: *** [all] Error 2
 >   > > >
 >   > > > The header files are there in:
 >   > > > /sw/include/freetype2/ftoutln.h
 >   > > >
 >   > > > So you can see that with the include paths
 >   > > > -I/sw/include -I/sw/include/freetype2
 >   > > >
 >   > > > it would not be able to see the path where the files actually
 >   > > > live. One can hack a fix by doing:
 >   > > > ln -s /sw/include/freetype2 /sw/include/freetype
 >   > >
 >   > > We're shaking out build issues related to the new freetype release.
 >   > > A number of other packages rely on the <freetype/HEADER.h>
 >   > > directory setup that you noted, and we’re fixing them accordingly.
 >   > >
 >   > > If folks do the hack, I’d recommend removing the symlink after your
 >   > > builds are done.
 >   >
 >   > But will this hack work in practice? I ask because in my earlier
 >   > attempts to convince somehow related things to build (namely,
 >   > <http://article.gmane.org/gmane.os.apple.fink.general/39311> and
 >   > <http://article.gmane.org/gmane.os.apple.fink.general/39313>) I tried
 >   > that first, and that caused the "Looking for incorrect headers in XXX
 >   > dependency files..." phase to complain loudly and kill the build
 >   > process. I am therefore wondering whether there is a trick to
 >   > overcome this issue.
 >   >
 >   > This is more a matter of curiosity than anything else though...
 >    
 >   One major pattern of bugs that the latest freetype (v2.5.something)
 > revealed is when a source hardcodes an #include for a header file named
 > freetype/*. That has been deprecated for a few years, but didn't become
 > functionally broken until now. Instead, sources are always supposed to
 > use macros for the headers. I pushed a fix for xft2 a few hours ago.
 > The symlink trick makes the older method work "sometimes" but may also
 > wind up pulling an incorrect header for some other part of some source
 > (which is what the fink-package-precedence check detects).

-- 
Stefan D. Bruda, PhD
Professor of Computer Science
Bishop's University 
2600 College St, Sherbrooke, Quebec J1M 1Z7, Canada
Phone: +1-819-822-9600 x2374, Fax: +1-819-822-9661
Web: <http://bruda.ca>, Email: <st...@br...>
PGP public key: <http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xD5DD6092>
* No HTML emails and proprietary attachments please <http://bruda.ca/ascii>
* Quid latine dictum sit altum videtur