1 PSPP is a program for statistical analysis of sampled data. It
2 is a free replacement for the proprietary program SPSS.
4 PSPP development is ongoing. It already supports a large subset of
5 SPSS's syntax. Its statistical procedure support is currently
6 limited, but growing. At your option, PSPP will produce statistical
7 reports in ASCII, PostScript, or HTML formats.
9 Instructions for PSPP installation are in INSTALL, including a list of
10 prerequisite packages and other PSPP-specific information. Full
11 documentation on PSPP's language will be installed along with the
14 For information on differences from previous versions, please see file
17 Source code for the latest release of PSPP is available at
18 ftp://ftp.gnu.org/pub/gnu/pspp/. Older versions may be obtained from
19 ftp://alpha.gnu.org/gnu/pspp/. Development sources are available
20 at http://savannah.gnu.org/projects/pspp
22 The following miscellaneous notes apply to this release:
24 * If you use GCC 4.3 or later to compile PSPP, along with any
25 version of the GNU Scientific Library (GSL) released as of
26 this writing (June 2008), then you will need to specify the
27 -fgnu89-inline flag to the compiler on the "configure"
28 command line: "./configure CFLAGS=-fgnu89-inline".
30 * On a few operating systems, such as OpenBSD, some of the
31 tests may fail with messages similar to: 'Warning: cannot
32 create a convertor for "646" to "UTF-8"'. These test
33 failures may safely be ignored.
35 Questions and comments about using PSPP may be sent to pspp-users@gnu.org.
36 Bug reports may be filed at http://savannah.gnu.org/bugs/?group=pspp
37 or emailed to bug-gnu-pspp@gnu.org. We prefer the web-based system
38 because it makes it more difficult for us to lose track of bugs, but we
39 are happy to hear from users through any means.
41 The long term goals for PSPP are ambitious. We wish to provide the
42 following support to users:
44 * All of the SPSS transformation language. PSPP already
45 supports a large subset of it.
47 * All the statistical procedures that someone is willing to
48 implement, whether they exist in SPSS or not.
50 * Compatibility with SPSS syntax, including compatibility with
51 known bugs and warts, where it makes sense. We also provide
52 an "enhanced" mode in certain cases where PSPP can output
53 better results that may surprise SPSS users.
55 * Friendly textual and graphical interfaces. This release
56 includes the first version of PSPPIRE, the PSPP graphical
59 * Attractive output, including graphs, in a variety of human-
60 and machine-readable formats. PSPP currently produces
61 output in ASCII, PostScript, and HTML formats. We will
62 enhance PSPP's output formatting in the future.
64 * Good documentation. Currently the PSPP manual describes its
65 language completely, but we would like to add information on
66 how to select statistical procedures and interpret their
69 * Efficient support for very large data sets. For procedures
70 where it is practical, we wish to efficiently support data
71 sets many times larger than physical memory. The framework
72 for this feature is already in place. It has not been tuned
73 or extensively tested, however initial experience has given
76 Over the long term, we also wish to provide support to developers who
77 wish to extend PSPP with new statistical procedures, by supplying the
80 * Easy-to-use support for parsing language syntax. Currently,
81 parsing is done by writing "recursive descent" code by hand,
82 with some support for automated parsing of the most common
83 constructs. We wish to improve the situation by supplying a
84 more complete and flexible parser generator.
86 * Easy-to-use support for producing attractive output.
87 Currently, output is done by writing code to explicitly fill
88 in table cells with data. We should be able to supply a
89 more convenient interface that also allows for providing
90 machine-readable output.
92 * Eventually, a plug-in interface for procedures. Over the
93 short term, the interface between the PSPP core and
94 statistical procedures is evolving quickly enough that a
95 plug-in model does not make sense. Over the long term, it
96 may make sense to introduce plug-ins.