PSPP NEWS -- history of user-visible changes.
-Copyright (C) 1996-9, 2000, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
+Copyright (C) 1996-2000, 2008-2016, 2017, 2019, 2020 Free Software Foundation, Inc.
See the end for copying conditions.
Please send PSPP bug reports to bug-gnu-pspp@gnu.org.
-Changes from 0.7.3 to 0.7.6:
+Changes from 1.4.0 to 1.4.1:
- * The "pspp" program has a new option --batch (or -b) that selects
- "batch" syntax mode. In previous versions of PSPP this syntax mode
- was the default. Now a new "auto" syntax mode is the default. In
- "auto" mode, PSPP interprets most syntax files correctly regardless
- of their intended syntax mode.
+ * Bug fixes.
- See the "Syntax Variants" section in the PSPP manual for more
- information.
+ * Translation updates.
- * The "pspp" program has a new option --syntax-encoding that
- specifies the encoding for syntax files listed on the command line,
- as well as the default encoding for syntax files included with
- INCLUDE or INSERT. The default is to accept the system locale
- encoding, UTF-8, UTF-16, or UTF-32, automatically detecting which
- one the system file uses.
+ * HTML output is now HTML5 instead of HTML4.0 Transitional.
- See the documentation for the INSERT command in the PSPP manual for
- more information.
+ * The HOST command has been reimplemented. It now supports TIMELIMIT.
+
+Changes from 1.2.0 to 1.4.0:
+
+ * The file pspp-mode.el (the pspp-mode for Emacs) is no longer distributed
+ with PSPP. Instead, it can be installed directly in emacs and is available
+ from https://elpa.gnu.org/packages/pspp-mode.html
+
+ * The Find dialog box, when searching for numeric values, will match only
+ to the precision of the variable's print format. This avoids behaviour
+ which is surprising to some users.
+
+ * PSPP now supports the SPSS viewer (.spv) format that SPSS 16 and later
+ use to save the contents of its output editor:
+
+ - PSPP and PSPPIRE can write output to .spv files.
+
+ - The new utility pspp-output can convert .spv files to other formats.
+
+ - The pspp-convert utility can now decrypt encrypted .spv files. The
+ encrypted viewer file format is unacceptably insecure, so to
+ discourage its use PSPP and PSPPIRE do not directly read or write
+ this format.
+
+ * A bug where the Data|Select Cases|Random Sample menu would generate
+ invalid syntax has been fixed.
+
+ * FREQUENCIES will emit a warning if one or more weight values are missing.
+
+ * Plain text output is no longer divided into pages, since it is now
+ rarely printed on paper.
+
+ * pspp-convert improvements:
+
+ - New "-a", "-l", "--password-list" options to search for an
+ encrypted file's password.
+
+ - New "--keep" and "--drop" options to output only selected variables.
+
+ - New "--recode", "--no-var-name", "--labels", "--print-formats",
+ "--decimal", "--delimiter", and "--qualifier" options to control
+ CSV output.
+
+ * Improvements to SAVE DATA COLLECTION support for MDD files.
+
+ * Bug fix for CVE-2018-20230.
+
+ * The /RENAME subcommand in SAVE TRANSLATE et al. has been extended
+ to support the syntax also accepted by other software.
+
+ * The EXAMINE command will now perform the Shapiro-Wilk test when
+ one or more plots are requested.
+
+ * The REGRESSION command now supports the /STATISTICS=TOL option which
+ outputs tolerance and variance inflation factor metrics for the data.
+
+ * The QUICK CLUSTER command now supports the /SAVE option which can
+ be used to save the cases' cluster membership and/or their distance
+ from the cluster centre to the active file.
+
+ * The MEANS command has been re-implemented.
+
+ * The GUI now has a sub-dialog for post-hoc tests of the oneway
+ anova test.
+
+ * Gtk+3.22.0 or later must now be used when building.
+
+ * GRAPH: Categorical barcharts displaying percentages now show the
+ sub-category as a percentage of the primary category instead of the
+ total.
+
+ * The Syntax Window now has a Find function to perform simple text
+ searching.
+
+ * AUTORECODE:
+
+ - The PRINT subcommand is now implemented.
+
+ - Many bug fixes.
+
+Changes from 1.0.1 to 1.2.0:
+
+ * New experimental command SAVE DATA COLLECTION to save MDD files.
+
+ * MTIME and YMDHMS variable formats now supported.
+
+ * Build changes:
+
+ - zlib is now a required dependency. (Previously it was optional.)
+
+ - Gtk+3.18.0 or later must now be used when building.
+
+ - The code implementing the sheet rendering has been removed. Instead we
+ use a third party library: spread-sheet-widget.
+
+ * The "NOMISSING" keyword when generating syntax for piecharts in the
+ frequencies command was omitted. This has been corrected.
+
+Changes from 1.0.0 to 1.0.1:
+
+ * Bug fixes, including fixes for CVE-2017-12958, CVE-2017-12959,
+ CVE-2017-12960, and CVE-2017-12961.
+
+Changes from 0.11.0 to 1.0.0:
+
+ * Translation updates.
+
+Changes from 0.10.2 to 0.11.0:
+
+ * The REGRESSION command now has a /ORIGIN subcommand to perform
+ regression through the origin.
+
+ * The FACTOR command can now analyse matrix files prepared with MATRIX DATA.
+
+ * The FACTOR command can now print the anti-image matrices.
+
+ * The MATRIX DATA command has been added.
+
+ * Some inappropriate properties in selection dialogs have been corrected.
+
+ * A bug which could cause the HTML driver to go into a tight loop
+ has been fixed.
+
+ * An error in the FREQUENCIES procedure, where the word "Mean" was
+ printed when "Variance" was appropriate has been fixed.
+
+ * The ncurses library is no longer required or used.
+
+ * A bug where the Mann-Whitney test would give misleading results
+ if run on multiple variables and MISSING=ANALAYSIS was specified
+ has been fixed.
+
+ * Gtk+3.14.5 or later must now be used when building.
+
+ * Graphical user interface changes:
+
+ ** There is a new menu: Edit|Options
+
+ ** The Non Parametric Statistics Menu has a new item: "K Independent Samples".
+
+ ** Dialog boxes can now be canceled using the <Escape> key.
+
+ * The AUTORECODE command now accepts an optional / before INTO.
+
+ * The short form of the VECTOR command can now create string variables.
+
+ * Bug fixes, including fixes for CVE-2017-10791 and CVE-2017-10792.
+
+Changes from 0.10.1 to 0.10.2:
+
+ * CROSSTABS implements a new COUNT subcommand to round case or cell
+ weights.
+
+ * Help will be opened as HTML in default browser if yelp
+ is not available.
+
+ * When plotting scatterplots with only one dataset (the simple case)
+ the colour used for the dataset is now black. The previous default
+ from the Tango palette was too faint to see easily.
+
+ * The varible info dialog could previously only paste a single variable.
+ Now it can paste multiple variables.
+
+ * Bug fixes, including the following:
+
+ - T-test with independent samples GUI crashed with string
+ type variable as group variable.
+
+ - The variable info dialog showed the previous selected variable
+ superimposed with the currently selected one.
+
+ - The GLM command did not properly deal with missing values. This
+ has been fixed.
+
+Changes from 0.10.0 to 0.10.1:
+
+ * Bug fixes, including the following:
+
+ - A bug where certain dialog boxes could not be properly populated
+ has been fixed.
+
+ - Various commands have been hardened against errors when called
+ with invalid syntax.
+
+ * Translation updates. Thanks to our translators!
+
+Changes from 0.8.5 to 0.10.0:
+
+ * The QUICK CLUSTER command has a /PRINT subcommand which shows
+ the initial cluster centres and the final cluster membership of
+ each case. The clustering algorithm has also been updated, so
+ as to produce better separated clusters.
+
+ * A Russian localisation has been contributed.
+
+ * The GRAPH command now has a /BAR subcommand to draw barcharts.
+
+ * If the DECIMAL character is set to COMMA then the ',' character
+ will not be treated as a separator by DATA LIST.
+
+ * The graphical user interface (psppire) has been changed as follows:
+
+ - It now uses Gtk+ version 3 instead of version 2. Accordingly, it has a
+ somewhat different look and feel.
- * The INCLUDE and INSERT commands now support the ENCODING subcommand
- to specify the encoding for the included syntax file.
+ - There is now a Graphs menu to access the GRAPH command.
- * Strings may now include arbitrary Unicode code points specified in
- hexadecimal, using the syntax U'hhhh'. For example, Unicode code
- point U+1D11E, the musical G clef character, may be expressed as
- U'1D11E'.
+ - The status of dialog box widgets are now preserved between calls
+ to the same dialog box for the same dataset.
- See the "Tokens" section in the PSPP manual for more information.
+ - The dialog box for the Logistic Regression command will now
+ infer that string variables or any varible with a "measure"
+ of Nominal or Ordinal are to be treated as categorical variables
+ and will generate syntax accordingly.
- * In previous versions of PSPP, in a string expressed in hexadecimal
- with X'hh' syntax, the hexadecimal digits expressed bytes in the
- locale encoding. In this version of PSPP, X'hh' syntax always
- expresses bytes in UTF-8 encoding.
+ * The pspp-convert utility can now decrypt encrypted syntax files.
+ The encrypted syntax file format is unacceptably insecure, so to
+ discourage its use PSPP and PSPPIRE do not directly read or write
+ this format.
- See the "Tokens" section in the PSPP manual for more information.
+ * New commands:
- * The DO REPEAT command has been reimplemented. The most prominent
- change is that when a DO REPEAT block contains an INCLUDE or INSERT
- command, substitutions are not applied to the included file.
+ - SORT VARIABLES.
- See the "DO REPEAT" section in the PSPP manual for more information.
+ * The following functions for transformation expressions are new:
- * NPAR TESTS now supports the /KRUSKAL-WALLIS and /RUNS subcommands.
+ - REPLACE, for search-and-replace of one string with another.
- * AUTORECODE now supports the /GROUP subcommand.
+ - STRUNC, to truncate a string and trim trailing spaces.
- * The MRSETS command is now implemented.
+ - MEDIAN, to compute the median of its arguments.
- * SAVE TRANSLATE is now implemented, with initial support for saving
- data in comma-separated value and tab-delimited formats.
+ - The TRUNC function in expressions now supports additional arguments
+ for truncating to values other than integers and to indicate a level
+ of rounding fuzz. The default rounding fuzz may now be controlled
+ and displayed with SET FUZZBITS and SHOW FUZZBITS, respectively.
- * The PRESERVE and RESTORE commands are now implemented.
+ * Bug fixes, including the following notable ones:
- * The HOST command has been updated to use more modern syntax.
+ - The correlation coefficient in the paired samples t-test
+ procedure is now correctly calculated when presented with
+ weighted data.
- * The SET and SHOW commands no longer support the ENDCMD, NULLINE,
- PROMPT, CPROMPT, and DPROMPT subcommands. The defaults are now the
- only supported values.
+ - The ESCAPE subcommand has been removed from GET DATA /TYPE=TXT.
+ For compatibility, the behavior that ESCAPE enabled is now the
+ default and only supported behavior.
-Changes from 0.7.2 to 0.7.3:
+ - Rendering of the variable and data sheets in right-to-left locales
+ now works properly.
- * Charts are now produced with Cairo and Pango, instead of libplot.
- Without them, the new graphing features will not work. If you do
- not have Cairo and Pango installed, you must run `configure' with
- --without-cairo.
+ - The Mann-Whitney test, when run on a dataset containing values
+ of the group variable, other than those defining the test groups,
+ would incorrectly calculate the ranks. This has been corrected.
- * The new "cairo" output driver supports output in PostScript, PDF,
- and SVG formats. Its functionality is a superset of that of the
- "postscript" driver, which has been removed. You must have Cairo
- and Pango installed to build the "cairo" driver.
+ * The IMPORTCASES subcommand on GET DATA is now ignored, for
+ compatibility.
-Changes from 0.7.1 to 0.7.2:
+Changes from 0.8.4 to 0.8.5:
- * Updated Perl module interface.
+ * The FREQUENCIES and CROSSTABS commands can now generate barcharts.
- * Value labels for long string variables are now supported.
+ * The FACTOR command can now perform PROMAX rotations.
- * Missing values for long string variables are now supported.
+ * SPSS/PC+ system files are now supported on GET and other commands
+ that read SPSS system files. The pspp-convert program can now read
+ SPSS/PC+ system files. Writing the obsolete SPSS/PC+ system file
+ format is not supported.
-Changes from 0.7.0 to 0.7.1:
+ * SYSFILE INFO can now read SPSS/PC+ system files and SPSS portable
+ files.
- * Added a perl module to facilitate reading and writing of pspp system
- files from perl programs.
+ * FREQUENCIES: A bug was fixed where an assertion failure occurred
+ when an empty dataset was presented.
-Changes from 0.6.2-pre6 to 0.7.0:
+ * The GRAPH command is now available. Initially it supports
+ scatterplots and histograms.
- * Custom variable and data file attributes are now supported.
- Commands VARIABLE ATTRIBUTE and DATAFILE ATTRIBUTE have been added
- for setting and clear attributes. Support for attributes has also
- been added to commands that read and write system files, such as
- SAVE and GET, as well as to the DISPLAY command.
+ * The RND operator in expressions now supports additional operands
+ for rounding to values other than integers and to indicate a level
+ of rounding fuzz. The default rounding fuzz may now be controlled
+ and displayed with SET FUZZBITS and SHOW FUZZBITS, respectively.
- * Numererous improvements to the Graphical User Interface have
- made. Notable improvements include:
+Changes from 0.8.3 to 0.8.4:
- - Non-Ascii characters in strings, labels and variable names are
- now supported.
+ * Formatting of SYSFILE INFO output was made easier to read.
- - A "Split Window" function is available, which makes it easier to
- see different parts of a large data file.
+ * Bug fixes, including the following notable ones:
- - Data files can now be opened by specifing their name as the first
- argument. This means that on a properly configured desktop, double
- clicking on an icon will open the file.
-
+ - FREQUENCIES works properly for string variables. (This bug was
+ introduced in 0.8.2.)
- * New statistical procedures:
- - CORRELATIONS
- - ROC
- - RELIABILITY
+ - CROSSTABS now correctly computes all of the measures that it
+ offers. Some measures have been removed because they were not
+ computed correctly.
- NPAR TESTS now supports the WILCOXON and SIGN subcommands.
+ - The NPAR TESTS calculation of significance for the RUNS
+ subcommand has been corrected.
- The CROSSTABS command has been completely re-implemented to fix numerous bugs.
+ - Planned comparisons in ONEWAY ANOVA now correctly handle negative
+ T-values.
- * Three new commands to combine data files have been added: MATCH FILES,
- UPDATE and ADD FILES.
+ - The COUNT command now correctly treats missing values as
+ documented.
- * A tutorial chapter has been added to the user manual.
+ - Conformance fixes to Open Document output format.
-Changes from 0.6.1 to 0.6.2
+Changes from 0.8.2 to 0.8.3:
+
+ * Independent T-Tests displayed the standard error of the difference
+ and the confidence interval for the unequal variances case, where
+ the equal variances case should have been. This has been fixed.
+
+ * REGRESSION now recognises /STATISTICS=CI(x) which causes confidence
+ intervals for the coefficients to be printed.
+
+ * When DESCRIPTIVES calculates Z scores, it now omits cases filtered
+ by the current FILTER settings.
+
+ * PSPPIRE graphical user interface improvements:
+
+ - Dialog boxes with source variable lists can now choose the sort
+ order; right click and select from the popup menu.
+
+ - File|Open now allows an encoding to be selected for opening
+ system files.
+
+ - File|Display Data File Information|External File... now allows an
+ encoding to be selected.
+
+ - A problem with the Means dialog has been resolved (bug #41433).
+
+ - Several problems related to the input of spreadsheets have been
+ fixed (bug #41657, #41620
+
+ * System file related improvements:
+
+ - With ENCODING="DETECT", SYSFILE INFO can now help the user to
+ pick an encoding for reading a system file that does not identify
+ its own encoding
+
+ - SYSFILE INFO now accepts an ENCODING subcommand to specify the
+ character encoding of string data in the system file.
+
+ - Variable labels over 255 bytes long are now accepted without
+ truncation (bug #41863).
+
+ - System files that contain duplicate variable names may now be
+ read successfully (bug #41475).
+
+Changes from 0.8.1 to 0.8.2:
+
+ * Charts are now rendered with colours from the Tango palette instead
+ of fully saturated primaries.
+
+ * Support for new system file variants:
+
+ - PSPP can now read and write ZCOMPRESSED system files, which
+ compress data much more effectively than older "compressed"
+ files. (The older format is still supported.)
+
+ - PSPP can now decrypt encrypted system files, using the new
+ pspp-convert utility. The encrypted system file format is
+ unacceptably insecure, so to discourage its use PSPP and PSPPIRE
+ do not directly read or write this format.
+
+ * Missing values for long string variables are now read from and
+ written to system files in an SPSS-compatible fashion.
+
+ (Earlier versions of PSPP that supported missing values for long
+ string variables wrote them to system files in an SPSS-incompatible
+ way. To fix the problem, read the system file with this version of
+ PSPP and then save a new copy of it.)
+
+ * pspp-convert, a new standalone utility for converting SPSS system
+ and portable files to other formats, is now included. The initial
+ version supports comma-separated value files as output format.
+ pspp-convert can also decrypt encrypted system files.
+
+ * Build changes:
+
+ - zlib is now a required dependency. (Previously it was optional.)
+
+Changes from 0.8.0 to 0.8.1:
+
+ * New commands:
+
+ - VARIABLE ROLE.
+
+ * Changes to existing commands:
+
+ - CROSSTABS: The requirement that one expected count must be less than five
+ before Fishers Exact Test will be reported has been removed.
+
+ * PSPPIRE graphical user interface improvements:
+
+ - A new set of icons has been contributed.
+
+ - Syntax windows now parse syntax in "auto" mode, which in practice
+ should mean that both "batch" and "interactive" syntax now works,
+ instead of just "interactive" syntax.
+
+ - The variable pane of data windows now include a Role column.
+
+ * Notable bug fixes:
+
+ - System files written by IBM SPSS 21 are now read without warnings.
+
+ - System files written by "VOXCO INTERVIEWER 4.3" are now read
+ without warnings.
+
+ - PSPPIRE should now more gracefully handle syntax files that contain
+ errors.
+
+Changes from 0.6.2 to 0.8.0:
+
+ * New commands:
+
+ - ADD FILES
+ - CORRELATIONS
+ - DATAFILE ATTRIBUTES
+ - DATASET ACTIVATE
+ - DATASET CLOSE
+ - DATASET COPY
+ - DATASET DECLARE
+ - DATASET DISPLAY
+ - DATASET NAME
+ - LOGISTIC REGRESSION
+ - MATCH FILES
+ - MEANS
+ - MRSETS
+ - PRESERVE and RESTORE
+ - QUICK CLUSTER
+ - RELIABILITY
+ - ROC
+ - SAVE TRANSLATE to CSV and tab-delimited files
+ - UPDATE
+ - VARIABLE ATTRIBUTES
+
+ * Changes to existing commands:
+
+ - AUTORECODE has a new GROUP subcommand.
+
+ - CROSSTABS has been re-implemented to fix numerous bugs.
+
+ - EXAMINE: /MISSING=LISTWISE is now the default.
+
+ - DO REPEAT command has been reimplemented. Now, when DO REPEAT
+ contains an INCLUDE or INSERT command, substitutions are not
+ applied to the included file.
+
+ - FILE HANDLE has a new ENDS subcommand to select CR or CRLF new-lines.
+
+ - HOST has been updated to use more modern syntax.
+
+ - Most commands that work with data files now support a new
+ ENCODING subcommand.
+
+ - MISSING VALUES can now assign missing values to long string
+ variables.
+
+ - ONEWAY: the POSTHOC subcommand is now implemented.
+
+ - The following new subcommands to NPAR TESTS have been implemented:
+ COCHRAN, FRIEDMAN, JONCKHEERE-TERPSTRA, KENDALL, KRUSKAL-WALLIS, MANN-WHITNEY,
+ MCNEMAR, SIGN, WILCOXON, and RUNS
+
+ - SET and SHOW no longer have ENDCMD, NULLINE, PROMPT, CPROMPT, and
+ DPROMPT subcommands. The defaults are now fixed values.
+
+ - SHOW now has a JOURNAL subcommand, to show the location of the
+ journal file.
+
+ - VALUE LABELS can now assign value labels to long string
+ variables.
+
+ * Other language changes:
+
+ - The new DATASET commands replace the "scratch file" PSPP
+ extension, which is no longer supported.
+
+ - Strings may now include arbitrary Unicode code points specified
+ in hexadecimal, using the syntax U'hhhh'. For example, Unicode
+ code point U+1D11E, the musical G clef character, may be
+ expressed as U'1D11E'.
+
+ See the "Tokens" section in the PSPP manual for more information.
+
+ - In previous versions of PSPP, in a string expressed in
+ hexadecimal with X'hh' syntax, the hexadecimal digits expressed
+ bytes in the locale encoding. In this version of PSPP, X'hh'
+ syntax always expresses bytes in UTF-8 encoding.
+
+ See the "Tokens" section in the PSPP manual for more information.
+
+ * PSPPIRE graphical user interface improvements:
+
+ - Added support for non-ASCII characters in strings, labels and
+ variable names.
+
+ - A "Split Window" function is available, which makes it easier to
+ see different parts of a large data file.
+
+ - Data files can now be opened by specifing their name as the first
+ argument. This means that on a properly configured desktop, double
+ clicking on an icon will open the file.
+
+ - Foreign formats, such as Gnumeric, OpenDocument and CSV can be imported
+ via using an "intelligent" application assistant dialog.
+
+ * Output changes:
+
+ - The new "cairo" output driver supports output in PostScript, PDF,
+ and SVG formats. Its functionality is a superset of that of the
+ "postscript" driver, which has been removed. You must have Cairo
+ and Pango installed to build the "cairo" driver.
+
+ - Charts are now produced with Cairo and Pango, instead of libplot.
+ Without them, the new graphing features will not work. If you do
+ not have Cairo and Pango installed, you must run `configure' with
+ --without-cairo.
+
+ - The plain text output driver now properly supports multibyte UTF-8
+ characters, including double-width characters and combining
+ accents.
+
+ - Output to OpenDocument format is now supported.
+
+ - The HTML output is much improved.
+
+ * The "pspp" program has a new option --batch (or -b) that selects
+ "batch" syntax mode. In previous versions of PSPP this syntax mode
+ was the default. Now a new "auto" syntax mode is the default. In
+ "auto" mode, PSPP interprets most syntax files correctly regardless
+ of their intended syntax mode.
+
+ See the "Syntax Variants" section in the PSPP manual for more
+ information.
+
+ * The "pspp" program has a new option --syntax-encoding that
+ specifies the encoding for syntax files listed on the command line,
+ as well as the default encoding for syntax files included with
+ INCLUDE or INSERT. The default is to accept the system locale
+ encoding, UTF-8, UTF-16, or UTF-32, automatically detecting which
+ one the system file uses.
+
+ See the documentation for the INSERT command in the PSPP manual for
+ more information.
+
+ * Text data files that PRINT and WRITE creates now use the system
+ native line ends by default (CRLF on Windows, LF only elsewhere).
+ Use the new ENDS subcommand on FILE HANDLE to override the default.
+
+ * A new Perl module allows Perl programs to read and write PSPP
+ system files.
+
+ * A tutorial chapter has been added to the user manual.
+\f
+Changes from 0.6.1 to 0.6.2:
* New translations: