usage: ./configure [ variable=value ]... This configure script generates the file 'makefile' and the file '../include/NTL/config.h', based upon the values assigned to the variables on the command line. Note that all of these configuration options can also be set by editing these two (well documented) files by hand. This command is intended only to provide a slightly more convenient and (perhaps more importantly) non-interactive way to do this. This script does not perform any 'magic', like finding out what the local C compiler is called, etc. If the defaults are not correct for your platform, you have to set an appropriate variable. ########### Here are the most important variables, and their default values. CC=gcc # The C compiler CXX=g++ # The C++ compiler CFLAGS=-O2 # C complilation flags CXXFLAGS=$(CFLAGS) # C++ compilation flags (by default, same as CFLAGS) PREFIX=/usr/local # Directory in which to install NTL library components NTL_STD_CXX=off # ISO Mode switch NTL_GMP_LIP=off # Switch to enable the use of GMP as primary # long integer package NTL_GMP_HACK=off # Switch to enable the use of GMP as supplemental # long integer package GMP_PREFIX=none # Directory in which GMP components have been installed ########## Here are more detailed description of these variables. ########## Basic compilation variables: CC=gcc # A C or C++ compiler, e. g., gcc, cc, xlc. # There are a few components written in C which may be compiled # under C++, but C compilers tend to generate slightly better code. # The C++ and C compilers (if different) must generate # compatible code. CXX=g++ # A C++ compiler, e.g., g++, CC, xlC CFLAGS=-O2 # Flags for the C compiler # Some useful flags: # -O2 -- recommended level of optimization # -g -- debugging # -mv8 -- needed with gcc/g++ to get the full instruction set # of a SparcStation # -qarch=ppc -- needed with xlc/C to get the full instruction set # of a PowerPC CXXFLAGS=$(CFLAGS) # Flags for the C++ compiler (usually the same as CFLAGS) # Some useful flags: # -+ -- needed with xlC to make it treat .c files as C++ files ########## Installation path: PREFIX=/usr/local # Set this to the directory in which you want NTL components to be # installed. When 'make install' is executed, the header # files are copied into $(PREFIX)/include/NTL, the library itself is # copied to $(PREFIX)/lib/libntl.a, and the documentation files # are copied into $(PREFIX)/doc/NTL. # Unless you have root permissions when running 'make install', # you will have to override the default PREFIX value with the # name of your own local directory. # If you want finer-grained control over where the different # library components are installed, set the variables # INCLUDEDIR, LIBDIR, and DOCDIR (see below). ########## ISO mode switch: NTL_STD_CXX=off # Set to 'on' if you want to use the "Standard C++" version of NTL. # In this version, all of NTL is "wrapped" inside the namespace NTL, # and are no longer directly accessible---you must either use # explicit qualification, or using directives, or using declarations. # However, note that all names that begin with "NTL_" are macros, # and as such do not belong to any namespace. # Additionally, instead of including the standard headers # , , and , the standard headers # , , and are included. # These "wrap" some (but not all) names in namespace std. # If your compiler is not yet up to date, but you want some # of the benefits of Standard C++, you might try the "partial Standard C++" # switches NTL_PSTD_NNS, NTL_PSTD_NHF, NTL_PSTD_NTN (see below). ########## GMP variables: NTL_GMP_LIP=off # Set to 'on' if you want to use GMP, the GNU Multi-Precision package, # as the primary long integer package. # This will typically yield significantly faster long integer arithmetic # compared to the traditional long integer package, and will yield somewhat # faster long integer arithmetic compared the traditinal package # supplemented by GMP (i.e., the NTL_GMP_HACK flag). NTL_GMP_HACK=off # Set to 'on' if you want to use GMP, the GNU Multi-Precision package, # as a supplemental long integer package. # This will typically yield significantly faster long integer arithmetic # compared to the traditional long integer package, while still maintaining # complete backward compataibility with code that uses the traditional # long integer package. # If you set either of the above two flags, note the following. # If you have installed GMP in a standard location, this is # all you have to do. Otherwise, if GMP is built, but not installed # in a standard place, you have to set either the variable GMP_PREFIX, # or alternatively, the the variables GMP_INC and GMP_LIB as well (see below). GMP_PREFIX=none # If GMP was installed in a standard system directory, e.g., /usr/local, # then do not set this variable. # Otherwise, if you want to use GMP and GMP was installed in # a directory , then set GMP_PREFIX=. # This works if the directory /include contains gmp.h # and /lib contains libgmp.a; otherwise, do not set # this variable, and set the variables GMP_INC and GMP_LIB below. ########### Examples: # If you are happy with all the default values: ./configure # Actually, the initially installed makefile and config.h files # already reflect the default values. # If your C/C++ compilers are called cc/CC: ./configure CC=cc CXX=CC # If GMP is installed in a standard system directory, and you want to use it: ./configure NTL_GMP_LIP=on # If GMP was installed in a non-standard directory foo: ./configure NTL_GMP_LIP=on GMP_PREFIX=foo # If you want to use the options -g and -O for compiling C and C++, # just execute ./configure "CFLAGS=-g -O" # Note the use of quotes to keep the argument in one piece. # If you want to use GMP as well as ISO mode: ./configure NTL_GMP_LIP=on NTL_STD_CXX=on ########### Here is a complete list of the remaining variables, ########### with their default values. These variables are pretty ########### esoteric, and you will probably never change their ########### default values. AR=ar ARFLAGS=ruv RANLIB=ranlib LDFLAGS= LDFLAGS_CXX=$(LDFLAGS) LDLIBS=-lm LDLIBS_CXX=$(LDLIBS) CPPFLAGS= LIBDIR=$(PREFIX)/lib INCLUDEDIR=$(PREFIX)/include DOCDIR=$(PREFIX)/include NTL_PSTD_NNS=off NTL_PSTD_NHF=off NTL_PSTD_NTN=off NTL_LONG_LONG_TYPE=long long NTL_CXX_ONLY=off NTL_CLEAN_INT=off NTL_CLEAN_PTR=off NTL_RANGE_CHECK=off NTL_X86_FIX=off NTL_NO_X86_FIX=off NTL_NO_INIT_TRANS=off WIZARD=on NTL_LONG_LONG=off NTL_AVOID_FLOAT=off NTL_TBL_REM=off NTL_AVOID_BRANCHING=off NTL_FFT_PIPELINE=off ########### Here is a more detailed description of these variables. ########### Further compilation variables: AR=ar # command to make a library ARFLAGS=ruv # arguments for AR RANLIB=ranlib # set to echo if you want to disable it completely LDFLAGS= # arguments for linker for C++ programs LDFLAGS_CXX=$(LDFLAGS) # libraries for linking C programs LDLIBS=-lm # libraries for linking C++ programs LDLIBS_CXX=$(LDLIBS) # libraries for linking C++ programs CPPFLAGS= # arguments for the C preprocessor ########### Details of the compilation process: # When a C file foo.c is compiled: $(CC) -I../include $(CPPFLAGS) $(CFLAGS) -c foo.c # When a C file foo.c is compiled and linked: $(CC) -I../include $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) -o foo foo.c $(LDLIBS) # When a C++ file foo.c is compiled: $(CXX) -I../include $(CPPFLAGS) $(CXXFLAGS) -c foo.c # When a C++ file foo.c is compiled and linked: $(CXX) -I../include $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS_CXX) \ -o foo foo.c $(LDLIBS_CXX) # When the library ntl.a is built: $(AR) $(ARFLAGS) ntl.a [ object files ]... $(RANLIB) ntl.a # If the ranlib command does not exist, everything will still function OK. ########### Further installation variables: LIBDIR=$(PREFIX)/lib INCLUDEDIR=$(PREFIX)/include DOCDIR=$(PREFIX)/doc # Where to install NTL. # Execution of 'make install' copies header files into $(INCLUDEDIR)/NTL, # copies the library itself to $(LIBDIR)/libntl.a, and copies the # documentation files into $(DOCDIR)/NTL. ########## Partial ISO modes NTL_PSTD_NNS=off # NTL namespace -- wraps NTL in a namespace NTL_PSTD_NHF=off # new header files -- use , etc., instead of , etc. NTL_PSTD_NTN=off # nothrow new -- use the nothrow version of new. # Any combination of these PSTD swtiches may be set -- setting them all # is equvalent to setting NTL_STD_CXX. ########### Basic Configuration Options: NTL_LONG_LONG_TYPE=long long # Name of double-word signed integer type. # This is a non-standard type, and is called 'long long' by many # compilers. MS C++ calls it '__int64'. NTL_CXX_ONLY=off # It is possible to compile everything using C++ only. # If you want to do this, CC and CXX should both be C++ compilers. # You may also want to set NTL_CXX_ONLY=on, which eliminates some # "C" linkage that is no longer necessary. # However, it should still work without it. # # This flag can be set independently of NTL_STD_CXX. # All functions that may have "C" linkage are never wrapped in # namespace NTL; instead, their names always start with "_ntl_", # and as such, they should not conflict with other global names. NTL_CLEAN_INT=off # Setting this to 'on' will disallow the use of some non-standard # integer arithmetic which may yield somewhat better performance. NTL_CLEAN_PTR=off # Setting this to 'on' will disallow the use of some non-standard # pointer arithmetic which may yield somewhat better performance. NTL_RANGE_CHECK=off # Setting this to 'on' will generate vector subscript range-check code. # Useful for debugging, but it slows things down of course. NTL_X86_FIX=off # Set to 'on' to force the "x86 floating point fix", # overriding the default behavior. # By default, NTL will apply the "fix" if it looks like it is # necessary, and if knows how to fix it. # The problem addressed here is that x86 processors sometimes # run in a mode where FP registers have more precision than doubles. # This will cause code in quad_float.c some trouble. # NTL can normally correctly detect the problem, and fix it, # so you shouldn't need to worry about this or the next flag. NTL_NO_X86_FIX=off # Set to 'on' to forces no "x86 floating point fix", # overriding the default behavior. NTL_NO_INIT_TRANS=off # When 'off', NTL uses a special code sequence to avoid # copying large objects in return statements. However, if your # compiler optimizes away the return of a *named* local object, # this is not necessary, and setting this flag to 'on' will result # in *slightly* more compact and efficient code. Although # the emeriging C++ standard allows compilers to perform # this optimization, I know of none that currently do. # Most will avoid copying *temporary* objects in return statements, # and NTL's default code sequence exploits this fact. ########## Performance Options: WIZARD=on # Set to 'off' if you want to bypass the wizard; otherwise, set to 'on'. # The wizard is a script that runs when NTL is built that sets the following # flags to 'optimize' performance on the current platform. NTL_LONG_LONG=off # For platforms that support it, this flag can be set to cause # the low-level multiplication code to use the type "long long", # which on some platforms yields a significant performance gain, # but on others, it can yield no improvement and can even # slow things down. # The variable NTL_LONG_LONG_TYPE can be defined to use a type name # other than "long long". # If you set NTL_LONG_LONG, you might also want to set # the flag NTL_TBL_REM. NTL_AVOID_FLOAT=off # On machines with slow floating point or---more comminly---slow int/float # conversions, this flag can lead to faster code. # If you set NTL_AVOID_FLOAT, you should probably also # set NTL_TBL_REM. # Note that at most one of NTL_LONG_LONG and NTL_AVOID_FLOAT may be set. # Also note that if NTL_DIRTY_INT is not set, and neither NTL_AVOID_FLOAT # nor NTL_LONG_LONG are set, the default behavior is the same as if # NTL_AVOID_FLOAT were set. NTL_TBL_REM=off # With this flag, some divisions are avoided in the # ZZ_pX multiplication routines. If you use the NTL_AVOID_FLOAT # or NTL_LONG_LONG flags, then you should probably use this one too. NTL_AVOID_BRANCHING=off # With this option, branches are replaced at several # key points with equivalent code using shifts and masks. # Recommended for use with RISC architectures, especially # ones with deep pipelines and high branch penalities. # This flag is becoming less helpful as newer machines # have much smaller branch penalties, but still may be worth a try. NTL_FFT_PIPELINE=off # If using NTL_AVOID_BRANCHING, you might want to try this as well. #This causes the FFT routine to use a software pipeline.