(2.01080
still fully supports only 2.1.7, but mostly works with 2.3.* too.)The next paragraph is an excerpt from documentation inside Pari.pm:
Package Math::Pari is a Perl interface to famous library PARI for
numerical/scientific/number-theoretic calculations. It allows
use of most PARI functions (>500) as Perl functions, and (almost)
seamless merging of PARI and Perl data.
The code is distributed with the same license as Perl (i.e., Artistic or or GPL); however, it uses GP/PARI code, which is purely GPL.
b) This module includes a part written in C; thus it is requires a fully
working Perl installation, including a C compiler which Perl knows about. Since prebuild configurations are typically broken (in many subtle ways), build is not supported unless your perl executable was build on your machine (or otherwise you show that you know what you are talking about ;-).
Given this, in most cases, this module can be installed as all the others:
perl -MCPAN -e shell
CPAN> install Math::Pari
or
perl -MCPAN -e 'install Math::Pari'
or (if you manually downloaded the archive, and change version numbers):
tar -zxfv Math-Pari-2.001200.tar.gz
cd Math-Pari-2.001200
perl Makefile.PL && make && make test && make install
Possible complications are addressed in the file INSTALL:
x) Not having Net::FTP installed and working, and not having
GP/PARI build directory "nearby" (as parent, kid or sibling);
x) Failing compile or configuration: wrong assembler
architecture, wrong paricfg.h (CLK_TCK etc.);
x) the rest of it...
IMPORTANT REMARK: If you try to compile with a different configuration, please run
make realclean
in between. Otherwise significant problems may arise. [A typical symptom: the symbol overflow cannot be found.]
c) How the tests work
Lowercase test scripts translate native GP/PARI tests from GP syntax to Perl syntax. You can get a better understanding of these translations by running the tests manually, as in
perl t/all.t
after install, or
perl -Mblib t/all.t
before install. Currently two subtests of t/nfields 'leak' PARI warnings. Until the test engine is taught to look-for/eat these warnings, they must appear.
d) Reporting bugs/problems (REQUIRED!!!!!!!!!!!!!!!!!!!!!!!!!)
0) I need to see the output of
perl Makefile.PL
perl -V
and the relevant messages from the `make' step "near the failure point".
cp libPARI/paricfg.h libPARI/paricfg.h-no-configure
make realclean
perl Makefile.PL Configure
and retest the bug you found. If this does not fix the bug, just mention this in your bug report. If this fixes the bug, please supply the result of
diff -u libPARI/paricfg.h-no-configure libPARI/paricfg.h
too.
2) Additionally, if this does not fix the bug, and your build is
architecture-specific (you saw a message similar to
...I will use assembler build of type 'ix86'.
during `perl Makefile.PL'), please also do
make realclean
perl Makefile.PL machine=none
retest your problem, and report whether it fixed the bug.
3) elliptic.t subtest 41 often fails on GP/PARI itself, so currently I
do not consider its failure to be a Math::Pari's bug.
4) If build succeeds, but test not, you can get higher-granularity reports
from test files by running them as (in bash/DOSISH, and in tcsh syntax):
perl -Mblib t/55_intnum.t 2>&1 > intnum-log
perl -Mblib t/55_intnum.t >& intnum-log
Please inspect and include the generated log file in your report too.
e) Documentation to this package is contained inside Pari.pm. After
installing it is accessible by standard means, like
perldoc Math::Pari
man Math::Pari
and possibly in other forms particular for your system.
A copy of documentation of the PARI library in POD format is provided as well, access it as Math::libPARI:
perldoc Math::libPARI
man Math::libPARI
Since many POD converters are too stupid/buggy to process these docs, an alternative Math::libPARI.dumb is provided; use as in
perldoc Math::libPARI.dumb
etc.