X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2FChangeLog;h=1fdf0d54e451195d1425d1fc135d04d346f1d66d;hb=bf6d99567a762b20bee4bb71147b6387c986bb90;hp=debb7021b3f1d6f2c228f5a7061a51311fb657db;hpb=00fc43143334ef109424b6e4d9d0a0917eac5563;p=pspp diff --git a/src/ChangeLog b/src/ChangeLog index debb7021b3..1fdf0d54e4 100644 --- a/src/ChangeLog +++ b/src/ChangeLog @@ -1,8 +1,116 @@ +Sun Apr 11 14:22:12 2004 Ben Pfaff + + Rework moments routines for improved numerical stability based on + Michael Kiefte's advice. Any bugs or remaining numerical problems + are still mine though. + + There is now a struct moments1 for use with one-pass moments. It + uses a provisional means algorithm as an attempt to improve + accuracy of higher moments. The older struct moments now only + handles two-pass moments. + + * aggregate.c: Use moments1 instead moments. + + * descript.c: Revert previous change, which is no longer needed + due to the moments revision. + + * moments.c: (calc_moments) New function for calculating variance, + skewness, kurtosis. + (moments_pass_one) Only accumulate weights bigger than zero. + (moments_calculate) Allow calculating the mean on pass one, others + require pass two. Implement in terms of calc_moments(). + (struct moments1) New structure. + (init_moments1) New function. + (moments1_clear) Ditto. + (moments1_create) Ditto. + (moments1_add) Ditto. + (moments1_calculate) Ditto. + (moments1_destroy) Ditto. + (cmd_debug_moments) Deal with `struct moments' or `struct + moments1' as requested by user. + +Sun Apr 11 14:21:55 2004 Ben Pfaff + + * Makefile.am (pspp_SOURCES): Remove debug.c. + + * debug.c: Removed. It was empty anyway. + +Fri Apr 9 20:04:49 2004 Ben Pfaff + + * descript.c (calc_descriptives): Fix assert failure when only + MOMENT_MEAN is needed. + +2004-04-09 Michael Kiefte + + * descript.c: + + fixed problem with parsing in match_statistic() causing + "DESCRIPTIVE STAT=MEAN." to barf. + + "MEAN" is now default if "SORT" given without specification. + + Fixed infinite loop with "DESCRIPT GIBBERISH=ALL." Parsing is + generally less forgiving of syntax errors: better to have it do + nothing and type it in again then to not know what it actually did + instead. + + z-score transformation now checks score for user-missing values + and checks std_dev for SYSMIS. + +2004-04-06 Michael Kiefte + + * aggregate.c, crosstabs.q, descript.c, dictionary.c, frequencies.q, levene.c, t-test.q, var.h: + Changed dict_get_case_weight() to accept an additional int * flag + to complain about system-missing, user-missing, zero, or negative + weights and updated existing functions to pass int * to + dict_get_case_weight(). + +2004-04-05 jmd + + * main.c: Fixed configuration problems with gsl + + * t-test.q: Fixed some problems encountered when compiling under Cygwin + +2004-04-03 blp + + * lexer.c, ChangeLog: + Fix infinite loop on comment at end of file, add test. + +2004-04-03 jmd + + * settings.h, var.h, ChangeLog, Makefile.am, cmdline.c, command.c, command.h, error.h, filename.c, frequencies.q, lexer.h, main.c, q2c.c, set.q: + Fixed the calculation of percentiles and added --syntax and --algorithm options + +Sat Apr 3 11:43:37 2004 Ben Pfaff + + * lexer.c: (lex_skip_comment) Handle end-of-file correctly (I + hope). + +Sat Apr 3 15:00:18 WST 2004 John Darrington + + * frequencies.q: Fixed the calculation of percentiles + + * Makefile.am: Added the --ansi flag and dealt with the + consequences. Added some entries to PSPP_sources so that + make distcheck would pass + + * cmdline.c: Added the --syntax and --algorithm options + + * q2c.c: Added an implicit /ALGORITHM subcommand to everything. + +Fri Apr 2 11:25:22 WAST 2004 John Darrington + + * t-test.q, levene.c, levene.h Converted t-test (incl levene) to + use the new multipass_split_... mechanism. + Wed Mar 31 22:36:22 2004 Ben Pfaff * frequencies.q: (calc_stats) Use moments data structure and calc_seskew(), calc_sekurt() functions. + * set.q main.c settings.h Added support for --syntax and --algorithm + options + Tue Mar 30 22:04:19 2004 Ben Pfaff * vfm.c: Had to get last call to multipass_split_output() inside