Use the highest number for x you found. Fresco will only build 2.6.3 and later. I'd tried gcc Version 2.5.8, 2.6.0, 2.6.2, 2.6.3 and 2.7.2. Current: 2.8.1
Since 2.6.3 the current source tree should be able to compile
with a little bit more Optimization:
#define DefaultCDebugFlags -O3 -fomit-frame-pointer
inside xf86site.def to overwrite the default -O2.
With 2.7.x you must specify
#define UsePosix YES inside xf86site.def.
This is necessary to build the sources successfully.
Versions prior to 2.7.0 could define it, but don't need it for a
clean build.
With 2.8.1 on Pentium CPU we can use special 586 Optimization. Try the following define inside your host.def.
#define DefaultGcc2i386Opt \
-O2 -march=pentium -mpentium -Wall -Wno-implicit-int \
-fno-strength-reduce
/* default with 2.8.1, so not really needed
-malign-loops=2 -malign-jumps=2 -malign-functions=2
*/
The needed g++ Libraries for use with g++ 2.x.x. As this is only necessary for Fresco, it isn't needed anymore since X11R6.1.
You could use the assembler and linker the assembler is most preferred,and the linker is needed at least if you want to link libFresco.a within a Program. Don't use strip and ar/ranlib, the first generates buggy binaries when stripping (at least on my machines) and the last requires the use of ranlib after creating an archive, this is not configured. Current: 2.8.1.0.15 (Used: as, ld, ar, strip)
Due to better memory usage we should use GNU's malloc library on systems where possible.
Enable #define UseGnuMalloc YES
inside xf86site.def or within the Linkkit site.def.
Enable and set #define GnuMallocLibrary
to your needs, if it isn't like the default
-L/usr/local/lib -lgmalloc.
This is the "original" inline-math package available at your favorite Linux Mirror.
Use #define UseInlineMath YES
inside host.def to enable it. Please note the changes section
what else to do, to use this package.