1 2008-05-15 Ben Pfaff <blp@gnu.org>
5 * data-out.c (output_infinite): Use gsl_isnan instead of isnan,
6 and gsl_isinf instead of isinf, as a stopgap measure for
7 portability until appropriate gnulib modules are available.
9 * por-file-writer.c (format_trig_double): Similarly, use
10 gsl_finite instead of finite.
12 2008-03-18 John Darrington <john@darrington.wattle.id.au>
14 * data-in.c: If category is custom currency, then use
15 numeric format for input.
17 2008-03-06 Ben Pfaff <blp@gnu.org>
19 * dictionary.c (var_name_is_insertable): New function.
20 (make_hinted_name): Don't accept variable names that match PSPP
21 keywords. Thanks to Jason Stover for reporting the problem.
23 2008-03-06 Ben Pfaff <blp@gnu.org>
25 * format-guesser.c (syntax): Require month names to be spelled out
26 as English words, so that single characters that happen to be
27 Roman numerals don't get detected as months. Thanks to John
28 Darrington for reporting this bug.
30 2008-03-04 Ben Pfaff <blp@gnu.org>
32 Patch #6441. Reviewed by John Darrington.
34 * format.c (fmt_fix): New function.
35 (fmt_fix_input): New function.
36 (fmt_fix_output): New function.
38 * format.def: Correct minimum width for DATETIME format. It was
39 7, should have been 17.
41 * automake.mk: Add new files.
43 * format-guesser.c: New file.
45 * format-guesser.h: New file.
47 2008-02-18 Ben Pfaff <blp@gnu.org>
49 Patch #6426. Reviewed by John Darrington.
51 * format.c (min_width): Renamed fmt_min_width and made public.
52 Updated all references.
53 (max_width): Renamed fmt_max_width and made public. Updated all
55 (max_decimals): Renamed fmt_max_decimals and made public. Updated
57 (var_create): Use the new functions for default variable
59 (var_default_formats): New function.
60 (var_default_measure): New function.
61 (var_default_alignment): New function.
63 * format.h (macro FMT_MAX_NUMERIC_WIDTH): New macro.
65 2008-02-09 Ben Pfaff <blp@gnu.org>
67 Add a couple of extensions to GET DATA TYPE=TXT. Patch #6412.
68 Thanks to John Darrington for review.
70 * data-in.c (data_in): Add new argument to designate the last
71 column of the data field being parsed, for use in error messages.
74 * data-parser (struct data_parser): New member `quote_escape'.
75 (data_parser_create): Initialize quote_escape.
76 (data_parser_set_quotes): New function.
77 (cut_field): Support escaped quotes.
78 (parse_delimited_span): Ditto.
79 (parse_delimited_no_span): Ditto.
81 * get-data.c (parse_get_txt): Support ESCAPE extension subcommand
82 in enhanced mode. Only support multiple quote characters in
85 2008-02-06 John Darrington <john@darrington.wattle.id.au>
87 psql-reader.c psql-reader.h: Read more than one tuple at
88 once. Fix bug reading a query which returns no data. Fix bug
89 when transformation followed a reader.
90 Ask the server for the number of records in the query, for the
93 2008-02-05 John Darrington <john@darrington.wattle.id.au>
95 psql-reader.c: So yesterday they release postgresql 8.3.0
96 which has money represented with 64 bits. They must get
99 2008-02-02 John Darrington <john@darrington.wattle.id.au>
101 psql-reader.c psql-reader.h: New files. Thanks to Ben Pfaff
102 for reviewing this code.
104 2008-02-02 Ben Pfaff <blp@gnu.org>
108 * sys-file-reader.c (read_variable_record): Allow missing values
109 to be specified on long string variables, but warn about them
110 because PSPP does not yet support them.
111 (read_extension_record): Ignore extension records 20 and 21, which
112 PSPP does not yet support.
113 (read_header): Fix error message when floating-point format cannot
116 2008-02-01 Ben Pfaff <blp@gnu.org>
118 Patch #6386. Thanks to John Darrington for review and for the
119 updates to gnumeric-reader.c.
121 * dictionary.c (make_hinted_name): New function.
122 (make_numeric_name): New function.
123 (dict_make_unique_var_name): New function.
125 * gnumeric-reader.c (devise_name): Removed.
126 (munge_name): Removed.
127 (gnumeric_open_reader): Use new function
128 dict_make_unique_var_name.
130 * short-names.c (set_var_short_name_suffix): Use new function
133 2008-01-19 John Darrington <john@darrington.wattle.id.au>
135 * settings.c settings.h: Moved static variables into a
136 single struct. Renamed functions to have a settings_ prefix.
138 2008-01-14 John Darrington <john@darrington.wattle.id.au>
140 * data-out.c (output_AHEX): Corrected number of bytes in
141 call to output_hex. Closes bug #22011
143 2008-01-02 John Darrington <john@darrington.wattle.id.au>
145 * variable.c variable.h: Replaced var_get_value_name with
146 var_append_value_name which doesn't use any static data.
147 Thanks to Ben for review.
149 2007-12-07 Ben Pfaff <blp@gnu.org>
153 * casegrouper.c (casegrouper_get_next_group): Cause a casegrouper
154 made from an empty casereader produce a casegrouper with no
155 groups, instead of one with one group that has no cases.
157 * casereader.c (casereader_is_empty): New function.
159 2007-12-06 Ben Pfaff <blp@gnu.org>
163 * sys-file-reader.c (read_display_parameters): Handle variable
164 display parameters record with only 2 data items per variable.
165 Reported by Guido Gay <gay@irer.it>.
167 2007-12-04 Ben Pfaff <blp@gnu.org>
169 * identifier.c (lex_id_match_n): New function.
170 (lex_id_match): Reimplement in terms of lex_id_match_n.
172 2007-11-24 Ben Pfaff <blp@gnu.org>
174 * automake.mk (src_data_libdata_a_SOURCES): Add val-type.h, to fix
177 2007-11-24 Ben Pfaff <blp@gnu.org>
179 Fix warning reported when reading back system files that include
180 very long string variables. Thanks to Guido Gay <gay@irer.it> for
183 * short-names.c (short_names_assign): Fix dumb typo, in which `i'
184 was written where `j' was meant.
186 2007-11-18 Ben Pfaff <blp@gnu.org>
188 Properly write variables that include a range of missing values to
189 system files. Thanks to Guido Gay <gay@irer.it> for reporting
192 * sys-file-writer.c (write_variable): Correctly calculate
193 n_missing_values field when writing variables that include a range
196 2007-11-10 Ben Pfaff <blp@gnu.org>
198 Cleanups and bug fixes devised while writing up documentation.
201 * automake.mk (src_data_libdata_a_SOURCES): Add new files.
203 * dict-class.c: New file.
204 (dict_class_from_id): Move here.
205 (dict_class_to_name): Move here.
207 * dict-class.h: New file.
208 (enum dict_class): Move here. Change from consecutive integers to
209 consecutive bits, to make testing for any of multiple values
210 easier. Add new DC_ALL constant.
212 * dictionary.c (struct dictionary): Change `case_limit' from
213 size_t to casenumber.
214 (dict_get_vars): Make final argument an enum dict_class.
215 (dict_get_vars_mutable): Ditto.
216 (dict_get_case_limit): Change return value to casenumber.
217 (dict_set_case_limit): Change final argument to a casenumber.
218 (dict_unset_split_var): Add assertion. Rephrase slightly.
219 (dict_set_label): Use xstrndup to simplify.
221 * format.c (fmt_step_width): AHEX format also needs 2-byte
223 (fmt_set_style): Simplify assertions.
225 * missing-values.c (mv_add_num_range): Rename mv_add_range.
226 Simplify implementation.
227 (mv_has_value): Simplify implementation.
228 (mv_pop_value): Remove the first value, not the last, to avoid
229 having GET followed by SAVE reverse the order of missing values.
230 (mv_peek_value): Rename mv_get_value. Simplify assertion.
231 (mv_has_range): Simplify implementation.
232 (mv_peek_range): Rename mv_get_range.
233 (can_resize_string): Removed.
234 (mv_is_resizable): Use value_is_resizable.
235 (mv_resize): Use value_resize.
237 * short-names.h (SHORT_NAME_LEN): Move here.
239 * val-type.h: New file, for definitions related to type and width
240 of abstract values. Before, these definitions were mixed among
241 those related to "union value" and those related to variables.
242 (macro SYSMIS): Move here.
243 (macro LOWEST): Move here.
244 (macro HIGHEST): Move here.
245 (macro MAX_STRING); Move here.
246 (enum val_type): New enum with values VAL_NUMERIC and VAL_STRING.
247 Replaces enum var_type that had values VAR_NUMERIC and VAR_STRING.
248 All references updated.
249 (val_type_is_valid): New function. Replaces var_type_is_valid.
250 All references updated.
251 (val_type_from_width): New function. Replaces
252 var_type_from_width. All references updated.
254 * value-labels.c (val_labs_copy): Renamed val_labs_clone. All
256 (val_labs_can_set_width): Use value_is_resizable.
257 (val_labs_add): Simply return false if the value labels set is too
258 wide, instead of having undefined behavior.
259 (val_labs_replace): Ditto.
260 (val_labs_replace): Ditto.
261 (val_labs_first): Set iterator to null if iteration is complete.
262 (val_labs_first_sorted): Ditto.
263 (val_labs_done): Become a no-op if the iterator is null.
265 * value.c (value_is_resizable): New function.
266 (value_resize): New function.
268 * variable.c (var_get_dict_class): New function.
270 * variable.h (macro LONG_NAME_LEN): Rename VAR_NAME_LEN. Update
273 2007-11-08 Ben Pfaff <blp@gnu.org>
275 * data-in.c: Make formatted data parsing locale-independent.
276 (parse_number): Use c_strtod instead of strtod, to avoid
277 locale-specific behavior.
280 2007-11-06 Ben Pfaff <blp@gnu.org>
282 Patch #6256: add support for binary, 360 file formats. Reviewed
285 * data-in.c (struct data_in): Add `encoding' member.
286 (data_in): Add `encoding' parameter, and re-encode the data passed
287 in where appropriate. Update all callers to pass it in.
288 (parse_A): Implement EBCDIC recoding wart described in manual.
289 (parse_AHEX): Implement EBCDIC recoding.
291 * data-out.c (data_out_legacy): New function.
292 (data_out): Make into a wrapper around data_out_legacy.
294 * file-handle-def.c (struct file_handle): New member `encoding'.
295 (fh_create_file): Set encoding.
296 (fh_default_properties): Set default encoding.
297 (fh_get_legacy_encoding): New function.
299 * file-handle-def.h (enum fh_mode): New modes FH_MODE_FIXED
300 (that replaces FH_MODE_BINARY), FH_MODE_VARIABLE,
301 FH_MODE_360_VARIABLE, FH_MODE_360_SPANNED.
302 (struct fh_properties): New member `encoding'.
304 2007-11-05 Ben Pfaff <blp@gnu.org>
306 Patch #6258. Reviewed by John Darrington.
308 * file-handle-def.c (fh_lock): Add comment that TYPE should be
309 marked with N_() in the caller. Added these markings to each
310 caller too. Should make i18n easier.
311 Suggested by Chusslove Illich <caslav.ilic@gmx.net>.
313 2007-11-03 Ben Pfaff <blp@gnu.org>
315 Allow output files to overwrite input files (bug #21280).
317 * file-handle-def.c: Separate locking of files for input and for
318 output, to allow both to take place at once. Also, distinguish a
319 file handle from the identity of the underlying file, because the
320 identity of a file changes over time and the file handle can't
321 represent two different identities.
322 (struct file_handle): Remove `next', `open_cnt', `deleted',
323 `type', `open_mode', `aux', `identity' members. Change `id' from
324 char array to char *. Add `ref_cnt' member.
325 (file_handle_from_ll) New function.
326 (file_handles) Removed.
327 (named_handles) New variable.
328 (fh_init) Initialize named_handles.
329 (fh_done) Remove name from all named_handles.
330 (free_handle) Updated for modified struct file_handle.
331 (unname_handle) New function.
332 (fh_ref) New function.
333 (fh_from_file_name) Removed.
334 (fh_unref) New function.
335 (fh_unname) New function.
336 (fh_from_id) Rewritten.
337 (create_handle) Updated for modified struct file_handle.
338 (fh_inline_file) Increment the handle's ref count.
339 (fh_create_file) Updated for modified struct file_handle.
340 (fh_create_scratch) Ditto.
345 (fh_is_open) Removed.
346 (fh_get_id) Updated for modified struct file_handle.
347 (fh_get_default_handle) Increment the handle's ref count.
348 (fh_set_default_handle) Handle ref counts.
349 (struct fh_lock) New structure.
350 (locks) New static var.
351 (fh_lock) New function.
352 (fh_unlock) New function.
353 (fh_lock_get_aux) New function.
354 (fh_lock_set_aux) New function.
355 (fh_is_locked) New function.
356 (make_key) New function.
357 (free_key) New function.
358 (compare_fh_locks) New function.
359 (hash_fh_lock) New function.
361 * file-handle-def.h (enum fh_access) New enum.
363 * file-name.c: Made file_identity the same in all supported
365 (struct file_identity): New `name' member.
366 (fn_get_identity): For a file that doesn't exist, get the
367 dev/inode of its directory plus its name. If even the directory
368 doesn't exist, just use its name. Merge the Windows
369 implementation into the Unix one.
370 (fn_compare_file_identities): Rewritten. Merge the Windows
371 implementation into the Unix one.
372 (fn_hash_identity): New function.
374 * make-file.c (struct replace_file): New structure.
375 (all_files): New static var.
376 (replace_file_start): New function.
377 (replace_file_commit): New function.
378 (replace_file_abort): New function.
379 (free_replace_file): New function.
380 (unlink_replace_files): New function.
382 * por-file-reader.c (struct pfm_reader): Add `lock' member.
383 (close_reader): Unlock file.
384 (pfm_open_reader): Lock file.
386 * por-file-writer.c (struct pfm_writer): Add fh_lock, replace_file
388 (pfm_open_writer): Lock file and prepare for its replacement.
389 (close_writer): Unlock file.
391 * scratch-handle.h (struct scratch_handle): Add unique_id so that
392 different generations of a scratch file can be distinguished.
394 * scratch-reader.c (scratch_reader_open): Verify that the file is
397 * scratch-writer.c (struct scratch_writer): Add `lock' and `dict',
398 remove scratch_handle member.
399 (scratch_writer_open): Lock handle. Prepare to replace handle
400 data, instead of doing it immediately.
401 (scratch_writer_casewriter_destroy): Replace handle data and
404 * sys-file-reader.c (struct sfm_reader): Add `lock' member.
405 (sfm_open_reader): Lock file.
406 (close_reader): Unlock file.
408 * sys-file-writer.c (struct sfm_writer): Add fh_lock, replace_file
410 (sfm_open_writer): Lock file and prepare for its replacement.
411 (close_writer): Unlock file.
413 2007-11-02 Ben Pfaff <blp@gnu.org>
415 * data-out.c (output_number): Use isfinite (from C99) instead of
416 gsl_isfinite, since we now have universal support for it in
418 (output_infinite): Ditto.
419 [!HAVE_ROUND] (round): Remove definition, since we now have a
420 replacement in gnulib.
422 2007-10-27 John Darrington <john@darrington.wattle.id.au>
424 * dictionary.c: Fixed bug in dict_clone, where the vardict.dict member
425 wasn't initialised correctly. Closes bugs #21330 and 21397
428 2007-10-08 Ben Pfaff <blp@gnu.org>
430 Fix bug #21280. Thanks to John Darrington for review.
432 * file-name.c (create_stream): New function.
434 * por-file-writer.c (pfm_open_writer): Use fh_open to open the
435 file handle before creating the file, to ensure that we don't
436 truncate a file that we're reading. Make code easier to read by
439 * sys-file-write.c (sfm_open_writer): Ditto.
441 2007-10-01 Ben Pfaff <blp@gnu.org>
443 Fix bug #21192. Thanks to John Darrington for review.
445 * casereader.c (casereader_read): Decrement case_cnt before
446 calling the casereader's "read" member function, so that we
447 interact properly with lazy_casereader.
449 * datasheet.c: Add regression test for above bug fix.
450 (clone_datasheet): New function.
451 (lazy_callback): New function.
452 (check_datasheet_casereader): New function.
453 (check_datasheet): Check datasheet contents are reported correctly
454 through an ordinary casereader and a lazy casereader.
455 (clone_model): Use clone_datasheet.
457 2007-09-24 Ben Pfaff <blp@gnu.org>
459 Patch #6210. Reviewed by John Darrington.
460 * settings.c: Drop "long view", which is not needed any longer.
461 (static var long_view): Removed.
462 (force_long_view): Removed.
463 (get_viewwidth): Removed.
464 (init_viewport): Removed long_view reference.
466 2007-09-19 John Darrington <john@darrington.wattle.id.au>
468 * settings.c settings.h: Changed viewport's length and width to be
469 owned by the user interface which uses the data library. This allows
470 better abstraction, and makes dynamically adjustable dimensions easier.
472 2007-09-18 Ben Pfaff <blp@gnu.org>
474 * procedure.c (proc_extract_active_file_data): New function.
476 * lazy-casereader.h: New file.
478 * lazy-casereader.c: New file.
480 * casereader.c (casereader_dynamic_cast): New function.
482 2007-09-14 Ben Pfaff <blp@gnu.org>
484 * dictionary.c (dict_clone): Copy case indexes from cloned
485 dictionary. Fixes bug #21061. Reviewed by John Darrington.
487 2007-09-13 John Darrington <john@darrington.wattle.id.au>
489 * value.c value.h (create_value): New function.
491 2007-09-12 Ben Pfaff <blp@gnu.org>
493 Make it clear that translator casereader and casewriter translate
494 functions are supposed to destroy their input case.
496 * casereader-translator.c (struct casereader_translator): Change
497 input case parameter of translate member function from const
498 struct ccase * to struct ccase *.
499 (casereader_create_translator): Ditto, for translate parameter.
501 * casewriter-translator.c (struct casewriter_translator): Ditto.
502 (casewriter_create_translator): Ditto.
504 2007-08-27 John Darrington <john@darrington.wattle.id.au>
506 * sys-file-reader.c (read_display_parameters): Force display width
507 to 8, if the sys file says 0 (like SPSS does).
509 2007-08-12 Ben Pfaff <blp@gnu.org>
511 * dictionary.c (dict_dump): New function.
513 2007-08-12 Ben Pfaff <blp@gnu.org>
515 Drop dict_compactor in favor of using the new struct case_map.
517 * dictionary.c (struct copy_map): Removed.
518 (struct dict_compactor): Removed.
519 (dict_make_compactor): Removed.
520 (dict_compactor_compact): Removed.
521 (dict_compactor_destroy): Removed.
523 * procedure.c (struct data_set): Change `compactor' member to be a
525 (proc_open): Use case_map_to_compact_dict instead of
527 (proc_casereader_read): Use case_map_execute instead of
528 dict_compactor_compact.
529 (proc_commit): Use case_map_destroy instead of
530 dict_compactor_destroy.
532 * scratch-writer.c (struct scratch_writer): Change `compactor'
533 member to be a struct case_map *.
534 (scratch_writer_open): Use case_map_to_compact_dict instead of
536 (scratch_writer_casewriter_write): Use case_map_execute instead of
537 dict_compactor_compact.
539 2007-08-12 Ben Pfaff <blp@gnu.org>
541 * automake.mk: Add case-map.c, case-map.h.
543 * case-map.c: New file.
545 * case-map.h: New file.
547 2007-08-12 Ben Pfaff <blp@gnu.org>
549 * dictionary.c (dict_compact_values): Don't delete scratch
550 variables as well as compacting case indexes. Update all callers.
551 (dict_get_compacted_value_cnt): Rename dict_count_values and
552 change interface. Update all callers.
553 (dict_get_compacted_value_cnt): Remove.
554 (dict_compacting_would_shrink): Remove.
555 (dict_compacting_would_change): Remove.
556 (dict_make_compactor): Add new parameter. Update all callers.
558 * procedure.c (proc_casereader_read): Use casewriter_get_value_cnt
559 instead of dict_count_values, changing an O(N) operation into
562 2007-08-12 Ben Pfaff <blp@gnu.org>
564 * casereader.c (casereader_read): Don't require cases read by a
565 casereader to be exactly the expected size: as long as they're big
568 2007-08-12 Ben Pfaff <blp@gnu.org>
570 Make casewriters keep track of the number of `union value's in
571 each case. This is useful for two reasons: casewriter_write can
572 then check that the case being written is large enough, and later
573 recipients of the casewriter can determine the size of the case.
575 * casewriter-translator.c (casewriter_create_translator): Add
578 * casewriter.c (struct casewriter): Add value_cnt member.
579 (casewriter_write): Check that the case passed in is big enough.
580 (casewriter_get_value_cnt): New function.
581 (casewriter_create): Add value_cnt parameter.
583 2007-08-09 Ben Pfaff <blp@gnu.org>
585 Fix bug reported by Jason Stover.
586 * settings.c: Move get_termcap_viewport and in particular the
587 #include for <curses.h> to the end of the file. curses.h
588 redefines bool on some systems (e.g. OpenBSD), which causes
589 disagreement between uses of bool before its inclusion and
591 Tested by Jason Stover.
593 2007-07-29 Ben Pfaff <blp@gnu.org>
595 Provisional fix for bug #18692 and bug #20161. Reviewed by John
598 * file-name.c (fn_open): Only pass "r" or "w" to popen as mode
599 argument (never "rb" or "wb") because SUSv3 says that only those
600 modes are defined, and glibc in fact rejects other modes.
602 Open portable files with fn_open so that they can be read from
603 pipes. Fix missing fh_close call to go along with fh_open.
604 Report an error if the file close reports an error.
605 * por-file-reader.c (close_reader): New function.
606 (por_file_casereader_destroy): Use close_reader.
607 (pfm_open_reader): Open file with fn_open.
609 2007-07-28 Ben Pfaff <blp@gnu.org>
611 Make PSPP able to read all the portable files I could find on the
612 web. Thanks to John Darrington for review. Bug #17620.
613 * por-file-reader.c (struct pfm_reader): New member `line_length'.
614 (error): Print file offset in hexadecimal.
615 (warning): New function.
616 (advance): Treat lines less than 80 bytes long as padded to 80
618 (pfm_open_reader): Call read_documents if we find an "E" record.
619 (convert_format): Convert invalid formats to the default format
620 instead of aborting reading the file.
621 (read_variables): Rename duplicate variable names instead of
622 aborting reading the file.
623 (read_value_label): Allow string variables of different widths to
624 be assigned value labels in the same record. Replace duplicate
625 value labels instead of aborting.
626 (read_documents): New function.
628 * por-file-writer.c (pfm_open_writer): Call write_documents if the
629 dictionary has documents.
630 (write_documents): New function.
632 2007-07-25 Ben Pfaff <blp@gnu.org>
634 Fix bugs related to bug #17213.
636 * settings.c: Use HAVE_LIBNCURSES instead of HAVE_LIBTERMCAP,
637 since the former is what config.h has. Include the needed ncurses
639 (static var echo) Rename to `do_echo' because the original name is
640 the same as an ncurses identifier.
641 (get_termcap_viewport) Use error instead of msg.
643 * file-name.c (fn_interp_vars): Fix interpolation of $VARS.
644 (fn_close): Don't close stdin, stdout, stderr.
646 2007-07-26 John Darrington <john@darrington.wattle.id.au>
648 * procedure.c procedure.h: Added callbacks which get invoked whenever
649 a dataset's transformation chain changes.
651 2007-07-24 Ben Pfaff <blp@gnu.org>
654 * sys-file-writer.c (write_variable_display_parameters): Use new
655 var_default_display_width function to choose display width of
656 segments after the first one in a given variable.
657 * variable.c (var_create): Use var_default_display_width to pick
658 new variable's display width.
659 (var_default_display_width): New function.
660 Reviewed by John Darrington.
662 2007-07-24 Ben Pfaff <blp@gnu.org>
665 * por-file-writer.c (write_variables): Write weight variable.
666 Reviewed by John Darrington.
668 2007-07-23 Ben Pfaff <blp@gnu.org>
670 Improvements to system file reader and writer.
672 First, move all detailed knowledge of very long strings into
673 sys-file-private.[ch], so that this nasty stuff can be isolated.
675 * sys-file-private.c (REAL_VLS_CHUNK): New macro.
676 (EFFECTIVE_VLS_CHUNK): New macro.
677 (min_int): New function.
678 (max_int): New function.
679 (sfm_width_to_bytes): Rewrite.
680 (sfm_width_to_octs): New function.
681 (sfm_segment_alloc_width): New function.
682 (sfm_segment_alloc_bytes): New function.
683 (sfm_segment_used_bytes): New function.
684 (sfm_segment_offset): New function.
685 (sfm_segment_effective_offset): New function.
686 (sfm_dictionary_to_sfm_vars): New function.
688 * sys-file-private.h (MIN_VERY_LONG_STRING): Removed.
689 (EFFECTIVE_LONG_STRING_LENGTH): Removed.
690 (struct sfm_var): New structure.
692 Next, improvements to the system file reader.
694 * sys-file-reader.h (struct sfm_read_info): Changed `case_cnt' to
695 type casenumber. Added `version_major', `version_minor',
698 * sys-file-reader.c (struct sfm_reader): Replaced `flt64_cnt' by
699 `oct_cnt'. Rename `vars', `var_cnt' to `sfm_vars', `sfm_var_cnt'.
700 Change `case_cnt' to type casenumber. Removed `has_vls'.
701 (struct sfm_var): Removed.
702 (sfm_open_reader): Don't warn on wrong case size if the file was
703 written by SPSS 13, which tends to get it wrong. Use
704 sfm_dictionary_to_sfm_vars.
705 (read_header): Always output system file info.
706 (read_variable_record): Simplify code for reading missing values.
707 (read_machine_int32_info): Save version numbers from system file
708 into info struct passed as new argument.
709 (read_long_string_map): Restructured to use new sys-file-private
711 (read_value_labels): Use size_overflow_p.
712 (sys_file_casereader_read): Get rid of distinction between fast
713 and slow paths. Use information provided by sys-file-primate's
714 struct sfm_var to simplify code.
715 (skip_whole_strings): New function.
716 (read_int32): Renamed read_int. Changed return value to int.
718 (read_flt64): Renamed read_float. Changed return value to
719 double. Updated all callers.
720 (int32_to_native): Removed. Changed callers to use
722 (flt64_to_double): Removed. Changed callers to use float_convert.
724 Finally, get rid of int32, flt64 terminology and types in system
725 file writer. The former wasn't very useful since a POSIX "int"
726 can hold the whole range of int32 and we generally didn't have a
727 need for it to be exactly-32-bits, just at-least-32-bits. The
728 latter was inconvenient because we had to assume that it could be
729 different from double and thereby convert special values SYSMIS,
730 HIGHEST, LOWEST to and from it in multiple places. Instead, now
731 we just use "int" and "double" in most places, and do conversions,
732 if necessary, very close to where we do I/O. This change meant
733 that the writer code couldn't represent records in the file as C
734 structs any longer, but that's no great loss. The code actually
735 seems to be more readable without them.
737 Simplify the compression buffering code: only buffer as much as
738 necessary, which is no more than eight 8-byte units at any given
741 * sys-file-writer.c (typedef flt64): Removed.
742 (macro second_lowest_flt64): Removed.
743 (struct sysfile_header): Removed.
744 (struct sysfile_variable): Removed.
745 (struct sfm_writer): Removed `needs_translation', `has_vls',
746 `flt64_cnt'. Changed `compress' to type bool and `case_cnt' to
747 type casenumber. Renamed `vars' to `sfm_vars', `var_cnt' to
748 `sfm_var_cnt'. Replaced `buf', `end', `ptr', `x', `y' for
749 compression buffering by `opcodes', `opcode_cnt', `data',
750 `data_cnt'. Renamed `var_cnt_vls' as `segment_cnt'.
751 (sfm_open_writer): Use sfm_dictionary_to_sfm_vars. Use simple
752 data writer functions instead of structures.
753 (calc_oct_idx): New function.
754 (write_header): Use simple data writer functions instead of
756 (write_format_spec): Renamed write_format. New argument.
757 (write_variable_continuation_records): New function.
758 (write_variable): Use simple data writer functions instead of
759 structures. Use write_variable_continuation_records. Write
760 entire very long string instead of requiring caller to understand
762 (write_value_labels): Use simple data writer functions instead of
764 (write_documents): Ditto.
765 (write_variable_display_parameters): Use sys-file-private
766 functions to simplify. Use simple data writer functions instead
768 (write_vls_length_table): Use simple data writer functions instead
770 (write_longvar_table): Ditto.
771 (write_rec_7_34): Break into new functions
772 write_integer_info_record, write_float_info_record. Use simple
773 data writer functions instead of structures.
774 (buf_write): Removed.
775 (append_string_max): Removed.
776 (ensure_buf_space): Removed.
777 (sys_file_casewriter_write): Get rid of the distinction between
778 fast and slow paths, which didn't seem to be too useful. Use new
779 functions write_case_uncompressed, write_case_compressed.
780 (put_instruction): Removed.
781 (put_element): Removed.
782 (write_compressed_data): Removed.
783 (close_writer): Use flush_compressed. Only write case count to
784 system file if it will fit in the field.
785 (write_case_compressed): New function.
786 (write_case_uncompressed): New function.
787 (flush_compressed): New function.
788 (put_cmp_opcode): New function.
789 (put_cmp_number): New function.
790 (write_int): New function.
791 (convert_double_to_output_format): New function.
792 (write_float): New function.
793 (write_value): New function.
794 (write_string): New function.
795 (write_bytes): New function.
796 (write_zeros): New function.
797 (write_spaces): New function.
799 Reviewed by John Darrington.
801 2007-07-22 Ben Pfaff <blp@gnu.org>
803 Don't try to write very long strings to portable files. The
804 format does not support it.
806 * por-file-writer.c (MAX_POR_WIDTH): New macro.
807 (pfm_open_writer): Limit output width to MAX_POR_WIDTH.
808 (write_format): Add arg to take width to resize format to.
809 (write_value): Limit width of value written to MAX_POR_WIDTH.
810 (write_variables): Limit width of variable and its output formats
812 Reviewed by John Darrington.
814 2007-07-22 Ben Pfaff <blp@gnu.org>
816 * sys-file-reader.c (read_variable_to_value_map): Use max_warnings
817 local variable instead of literal 5.
818 Reviewed by John Darrington.
820 2007-07-22 Ben Pfaff <blp@gnu.org>
822 Fix problems with uniqueness of short names in system files with
823 very long string variables. Now a variable may have multiple
826 * automake.mk (src_data_libdata_a_SOURCES): Add new files
827 short-names.c, short-names.h.
829 * dictionary.c (dict_clone): Clone all the short names.
830 (compare_strings): Move into short-names.c.
831 (hash_strings): Ditto.
832 (set_var_short_name_suffix): Ditto.
833 (dict_assign_short_names): Ditto, rename short_names_assign,
834 change to assign all short names.
836 * por-file-writer.c (write_variables): Use short_names_assign
837 instead of dict_assign_short_names.
839 * short-names.c: New file.
841 * short-names.h: New file.
843 * sys-file-private.c (sfm_width_to_segments): New function.
845 * sys-file-reader.c (read_long_var_name_map): Save and restore all
846 the short names, not just the first one.
848 * sys-file-writer.c (cont_var_name): Removed.
849 (sfm_open_writer): Use short_names_assign instead of
850 dict_assign_short_names. Use unique short names assigned by
851 short_names_assign instead of those generated by cont_var_name.
853 * variable.c (struct variable): Remove `short_name' member,
854 replace by `short_names' and `short_name_cnt'.
855 (var_create) Initialize new members.
856 (var_get_short_name_cnt): New function.
857 (var_get_short_name): Now takes an index argument. Changed most
859 (var_set_short_name): Ditto.
860 (var_clear_short_name): Renamed var_clear_short_names, changed to
861 clear all short names.
863 Reviewed by John Darrington.
865 2007-07-22 Ben Pfaff <blp@gnu.org>
867 * variable.c (var_set_width): Use new var_set_width function.
869 * missing-values.c (mv_n_values): Drop assertion, which was not
872 * format.c (fmt_default_for_width): New function.
873 (fmt_resize): New function.
875 Reviewed by John Darrington.
877 2007-07-18 John Darrington <john@darrington.wattle.id.au>
879 * datasheet.c (datasheet_delete_columns): Added assertion to check
880 we're not deleting outside the range of the sheet.
883 * dictionary.c dictionary.h variable.c: Added the ability for string
884 variables to be resized.
886 * vardict.h: Added some prototypes (moved from dictionary.h) as
887 these should only be called by variable.c
890 2007-07-14 John Darrington <john@darrington.wattle.id.au>
892 * sfm-reader.c: Respect case_cnt field in file header.
894 2007-07-01 John Darrington <john@darrington.wattle.id.au>
896 * transformation.c transformation.h (trns_chain_execute): Changed the
897 signature (Patch #6057)
899 2007-06-10 Ben Pfaff <blp@gnu.org>
901 * casereader-filter.c (casereader_filter_destroy): Make sure to
902 write all the remaining excluded cases to the casewriter, if any.
904 * caseinit.c (init_list_destroy): Rewrite.
905 (init_list_clear): Ditto.
907 * casegrouper.c (casegrouper_get_next_group): Always set *reader
908 to null when returning false.
910 2007-06-06 Ben Pfaff <blp@gnu.org>
912 Actually implement the new procedure code and adapt all of its
913 clients to match. Also adapt all of the other case sources and
914 sinks in the tree and their clients to use the
915 casereader/casewriter infrastructure.
917 * automake.mk: Add and remove files.
919 * any-reader.c: Change into a casereader.
920 * por-file-reader.c: Ditto.
921 * scratch-reader.c: Ditto.
922 * sys-file-reader.c: Ditto.
924 * any-writer.c: Change into a casewriter.
925 * por-file-writer.c: Ditto.
926 * scratch-writer.c: Ditto.
927 * sys-file-writer.c: Ditto.
929 * procedure.c: Change to use casereader, casewriter, caseinit, and
930 other new infrastructure.
932 * scratch-handle.c: Adapt to new infrastructure.
934 * case-sink.c: Removed, now dead code.
935 * case-sink.h: Ditto.
936 * case-source.c: Ditto.
937 * case-source.h: Ditto.
938 * casefile-factory.c: Ditto.
939 * casefile-private.h: Ditto.
942 * casefilter.c: Ditto.
943 * casefilter.h: Ditto.
946 * fastfile-factory.c: Ditto.
947 * fastfile-factory.h: Ditto.
948 * storage-stream.c: Ditto.
949 * storage-stream.h: Ditto.
951 2007-06-06 Ben Pfaff <blp@gnu.org>
955 * automake.mk: Add new files.
957 * datasheet.c: New file.
959 * datasheet.h: New file.
961 2007-06-06 Ben Pfaff <blp@gnu.org>
963 Until now, the procedure code has provided a case to the
964 case_source, which has filled in the data values that come from
965 the active file. "Left" data values that don't come from the
966 active file naturally stay the same from case to case, because the
967 procedure code keeps using that same case.
969 One of the compromises that comes with the new procedure code is
970 that the active file allocates and provides its own case, which
971 the procedure code then has to resize to provide room for any
972 other variables that should go in the case and then fill in the
973 values of "left" variables. Then, when we're done with that case,
974 we have to save the values of "left" variables to copy into the
975 next case read from the active file.
977 The caseinit code helps with this.
979 * automake.mk: Add new files.
981 * caseinit.c: New file.
983 * caseinit.h: New file.
985 2007-06-06 Ben Pfaff <blp@gnu.org>
987 * value.h (value_cnt_from_width): New function.
989 * variable.c (var_get_value_cnt): Use new function.
991 2007-06-06 Ben Pfaff <blp@gnu.org>
993 Add casegrouper, to allow cases read from a given casereader to be
994 broken into groups, each of which has its own casereader.
995 Generally cases are grouped based on having equal values for some
998 * automake.mk: Add new files.
1000 * casegrouper.c: New file.
1002 * casegrouper.h: New file.
1004 2007-06-06 Ben Pfaff <blp@gnu.org>
1006 Add interface to lexicographical ordering of cases.
1008 * automake.mk: Add new files.
1010 * case-ordering.c: New file.
1012 * case-ordering.h: New file.
1014 2007-06-06 Ben Pfaff <blp@gnu.org>
1016 Add casereaders and casewriters, the basis of the new data processing
1017 implementation. A casereader is a uniform interface to reading cases
1018 from a data source; a casewriter is a uniform interface to writing
1019 cases to a data sink.
1021 * automake.mk: Add new files.
1023 * casereader-filter.c: New file.
1025 * casereader-provider.h: New file.
1027 * casereader-translator.c: New file.
1029 * casereader.c: New file.
1031 * casereader.h: New file.
1033 * casewriter-provider.h: New file.
1035 * casewriter-translator.c: New file.
1037 * casewriter.c: New file.
1039 * casewriter.h: New file.
1041 2007-06-06 Ben Pfaff <blp@gnu.org>
1043 "casewindow" data structure that extends the deque (from libpspp)
1044 of cases with the ability to dump cases to disk if we get too many
1047 * automake.mk: Add new files.
1049 * casewindow.c: New file.
1051 * casewindow.h: New file.
1053 2007-06-06 Ben Pfaff <blp@gnu.org>
1055 sparse_cases data structure that augments a sparse_array of cases
1056 with the ability to dump cases to disk if we get too many cases in
1059 * automake.mk: Add new files.
1061 * sparse-cases.c: New file.
1063 * sparse-cases.h: New file.
1065 2007-06-06 Ben Pfaff <blp@gnu.org>
1067 Adds a low-level on-disk case array data structure.
1069 * automake.mk: Add new files.
1071 * case-tmpfile.c: New file.
1073 * case-tmpfile.h: New file.
1075 2007-06-06 Ben Pfaff <blp@gnu.org>
1077 In a couple of places we calculate the maximum number of cases to
1078 keep in memory based on the user-defined workspace. Enable
1079 centralizing the calculation through a new function.
1081 * settings.c (get_workspace_cases): New function.
1083 2007-06-06 Ben Pfaff <blp@gnu.org>
1085 The casenumber type is defined in transformations.h, but case.h is
1086 a more sensible place. Move it.
1088 * case.h (CASENUMBER_MAX): New macro.
1089 (typedef casenumber): Move here, from transformations.h.
1091 2007-06-03 Ben Pfaff <blp@gnu.org>
1093 Slightly generalize case_to_values and case_from_values functions.
1095 * case.c (case_to_values): Rename case_copy_out, change interface.
1096 (case_from_values): Rename case_copy_in, change interface.
1098 * fastfile.c (fastfilereader_get_next_case): Update caller.
1099 (write_case_to_disk): Ditto.
1101 2007-06-02 Ben Pfaff <blp@gnu.org>
1103 Clean up after a forgotten part of patch #5829.
1105 * casedeque.h: Remove unused file.
1107 * automake.mk: Remove casedeque.h from sources.
1109 2007-05-10 Jason Stover <jhs@math.gcsu.edu>
1111 * category.c: Removed redundant #include
1113 2007-05-06 Ben Pfaff <blp@gnu.org>
1115 Abstract the documents within a dictionary a little better.
1116 Thanks to John Darrington for suggestion, initial version, and
1117 review. Patch #5917.
1119 * dictionary.c (struct dictionary): Change `documents' member from
1120 char * to struct string.
1121 (dict_clear): Destroy struct string.
1122 (dict_get_documents): Convert struct string to char *.
1123 (dict_set_documents): Set struct string. Pad to 80-character
1125 (dict_clear_documents): New function.
1126 (dict_add_document_line): New function.
1127 (dict_get_document_line_cnt): New function.
1128 (dict_get_document_line): New function.
1130 * dictionary.h (macro DOC_LINE_LENGTH): New macro.
1132 * sys-file-reader.c (read_documents): Use new document functions.
1134 2007-04-19 John Darrington <john@darrington.wattle.id.au>
1136 * sys-file-reader.c: When reading a system file which has no
1137 long name table, automatically create one where the long names
1138 are the lower case versions of the short names.
1140 2007-04-22 Ben Pfaff <blp@gnu.org>
1142 * dictionary.c (dict_set_split_vars): dict_destroy expects that
1143 dict_clear will free most data related to the dictionary.
1144 dict_clear does a decent job, except that dict_set_split_vars on
1145 some systems won't actually free the dict's "split" member.
1146 Instead, it'll allocate a 1-byte region. Fix this.
1148 * value.c (value_copy): New function.
1149 (value_set_missing): Ditto.
1151 2007-04-22 John Darrington <john@darrington.wattle.id.au>
1153 * Deleted existing category.h and moved cat-routines.h into
1154 category.h Encapsulated struct cat_vals better.
1156 2007-04-19 John Darrington <john@darrington.wattle.id.au>
1158 * sys-file-reader.c: When reading a system file which has no
1159 long name table, automatically create one where the long names
1160 are the lower case versions of the short names.
1162 2007-04-16 John Darrington <john@darrington.wattle.id.au>
1164 * sys-file-reader.c: Some versions of Other Software seem to
1165 produce system files with string variables' measure set to
1166 zero. We'll assume these are supposed to be nominal variables.
1168 2007-03-30 Ben Pfaff <blp@gnu.org>
1170 * procedure.c: Adapt to new deque data structure.
1172 Mon Feb 19 10:53:21 2007 John McCabe-Dansted <gmatht@gmail.com>
1173 Ben Pfaff <blp@gnu.org>
1175 * file-name.c: Mingw compatibility fixes.
1176 (fn_search_path): Use ISSLASH instead of comparing against '/'
1178 (fn_dir_name): Use dir_name from gnulib.
1179 (fn_is_absolute): Use IS_ABSOLUTE_FILE_NAME from gnulib.
1180 (fn_get_identity): Use GetFullPathName instead of canonicalize
1181 from gnulib, because the latter does not fully support
1182 Windows-style path names. Use this implementation based on the
1183 detected presence of Windows instead of the absence of Unix, since
1184 the new implementation is Windows-specific.
1185 (fn_compare_file_identities): In Windows implementation, compare
1186 names case-insensitively.
1188 Sun Feb 18 13:28:02 2007 Ben Pfaff <blp@gnu.org>
1190 * make-file.c: Don't include mkstemp.h, because gnulib now causes
1191 <stdlib.h> to have the same effect.
1193 Sun Feb 18 11:20:24 2007 Ben Pfaff <blp@gnu.org>
1195 * por-file-reader.c: Add missing _() around messages.
1197 Sun Feb 11 20:44:13 2007 Ben Pfaff <blp@gnu.org>
1199 * make-file.c: Include "mkstemp.h", without which linking on
1202 Thu Feb 8 14:59:05 2007 Ben Pfaff <blp@gnu.org>
1204 Reduce platform dependence.
1206 * file-name.c (fn_tilde_expand): Removed, and removed calls to it.
1207 Everywhere we using this, we really should have just depended on
1208 the shell to expand tildes.
1209 (fn_search_path): Simplify, given that we don't do tilde expansion
1211 (fn_normalize): Removed. Caller changed to use the canonicalize
1213 (fn_get_cwd): Removed. Only user was fn_normalize.
1214 (fn_is_absolute): Really only test for absolute names.
1215 (fn_is_special): Use pipe files if HAVE_POPEN, not if we're in
1217 (fn_readlink): Removed, as it was only used fn_normalize.
1218 (fn_exists): Assume the stat function is available; gnulib does.
1219 (fn_open): Use pipe files if HAVE_POPEN, not if we're in unix.
1221 Sat Feb 3 21:52:17 2007 Ben Pfaff <blp@gnu.org>
1223 * dictionary.c (dict_create_vector_assert): New function.
1225 Wed Feb 7 21:25:15 2007 Ben Pfaff <blp@gnu.org>
1227 * file-name.c (fn_normalize): Correct name of function
1228 fn_is_special. Thanks to John McCabe-Dansted <gmatht@gmail.com>
1229 for pointing this out.
1231 Thu Feb 1 16:53:37 2007 Ben Pfaff <blp@gnu.org>
1233 We are using a single member in struct file_handle, the "name"
1234 field, for more than one purpose. When it begins with '"', it's a
1235 file name; otherwise, it's a token that can be used to identify
1236 it. When that assertion fires, it's because we searched for the
1237 name case-sensitively as a file name (so that there was no match),
1238 and then we try to insert it case-insensitively as a token, which
1239 fails because duplicates aren't allowed.
1241 Solution: break the two purposes into two separate fields. This
1242 fixes the problem and likely makes the code easier to read too.
1244 Fixes bug #18922. Thanks to John Darrington for bug report and
1247 * file-handle-def.c (struct file_handle): New `id' member.
1248 (fh_from_name): Rename fh_from_id. Update all callers.
1249 (create_handle): New `id' parameter. Update all callers.
1250 (fh_create_file): Ditto.
1251 (fh_get_id): New function.
1253 Mon Jan 15 16:18:10 2007 Ben Pfaff <blp@gnu.org>
1255 * case.c (case_is_null): Change return type to bool.
1257 Mon Jan 15 10:57:28 2007 Ben Pfaff <blp@gnu.org>
1261 * case.c (case_clone) [DEBUGGING]: When debugging, don't use
1262 reference counting to share data. This makes it easy for
1263 valgrind, etc. to find accesses to cases that have been destroyed
1264 but have been kept around by another user's ref-count. This often
1265 happens when the data set is small enough to find in memory; if a
1266 bigger data set that would overflow to disk were used, then data
1267 corruption would occur.
1269 Mon Jan 15 10:55:18 2007 Ben Pfaff <blp@gnu.org>
1273 * case.c (case_unshare): Make it check internally whether the
1274 ref_cnt is greater than 1, so that the callers don't have to.
1275 Update callers not to check.
1277 Mon Jan 15 10:53:01 2007 Ben Pfaff <blp@gnu.org>
1279 Before, I was thinking that I might want to get rid of reference
1280 counting at some point. Now, I'm pretty sure that it's here to
1281 stay. Thus, because we have to store the value_cnt anyway for
1282 reference-counted cases, we might as well expose it to users.
1284 * case.c (case_get_value_cnt): New function.
1285 (case_resize): Drop OLD_CNT argument. Update all callers. Only
1286 resize case if its size actually changed.
1288 * casefile.c (casefile_append_xfer): Use case_get_value_cnt
1289 instead of peeking inside struct case directly.
1290 (casefile_append): Ditto.
1292 Mon Jan 15 10:50:22 2007 Ben Pfaff <blp@gnu.org>
1294 Get rid of the inlines for the case functions, which made the
1295 header file hard to read. (Also, in testing with "-O2 -DNDEBUG",
1296 the inlines didn't speed up "make check" at all, which is not a
1297 perfect benchmark but seems indicative.)
1299 * case.c: Remove #ifdef DEBUGGING...#endif around many function
1300 definitions. Remove some assertions on nonnull pointers that were
1301 redundant with a pointer dereference soon after in the function.
1303 (struct case_data): Move definition here from case.h.
1307 (case_data_wr): Ditto.
1309 Sun Jan 14 21:41:12 2007 Ben Pfaff <blp@gnu.org>
1311 * automake.mk: Add casedeque.h to sources.
1313 * casedeque.h: New file.
1315 * procedure.c: (struct dataset) Change lag_count, lag_head,
1316 lag_queue member into single struct casedeque member. Update all
1317 users to use the casedeque instead.
1320 Sun Jan 14 21:43:12 2007 Ben Pfaff <blp@gnu.org>
1322 * procedure.c: Simplify lagged cases interface. Updated all
1323 clients--well, the only client--to use the simplified interface.
1324 (dataset_n_lag) Removed.
1325 (dataset_set_n_lag) Removed.
1326 (dataset_need_lag) New function.
1328 Tue Jan 9 07:20:05 WST 2007 John Darrington <john@darrington.wattle.id.au>
1330 * dictionary.c procedure.c: More changes to ensure that callbacks occur
1331 whenever appropriate, but only when the dataset/dictionary is in a
1334 Sun Jan 7 08:33:04 WST 2007 John Darrington <john@darrington.wattle.id.au>
1336 * dictionary.c dictionary.h : Added callbacks for change of filter and
1337 split variables. Refactored some code to ensure that callbacks get
1338 invoked when appropriate.
1340 * procedure.c (proc_cancel_temporary_transformations): Make sure that
1341 replace_dict callback occurs when permanent_dict replaces the current
1344 Wed Jan 3 11:02:11 WST 2007 John Darrington <john@darrington.wattle.id.au>
1346 * dictionary.c dictionary.h : Added callback for when the weight
1347 variable of a dictionary changes.
1349 Mon Jan 1 10:36:26 WST 2007 John Darrington <john@darrington.wattle.id.au>
1351 * dictionary.c dictionary.h : Added replace_source and replace_dict
1352 callbacks, and functions to deal with them.
1354 Fri Dec 22 13:56:08 2006 Ben Pfaff <blp@gnu.org>
1356 Simplify missing value handling.
1358 * missing-values.h (enum mv_class): New type.
1359 (enum mv_type): Moved definition into missing-values.c and renamed
1360 each MV_* to MVT_*, to distinguish them from the exposed mv_class
1361 enums. Updated all uses.
1362 (struct missing_values): Changed type of `type' from `enum
1363 mv_type' to `int' because the definition is no longer exposed.
1365 * missing-values.c (mv_is_value_missing): Add new enum mv_class
1366 parameter. Update all callers.
1367 (mv_is_num_missing): Ditto.
1368 (mv_is_str_missing): Ditto.
1369 (mv_is_value_user_missing): Removed. Changed callers to use
1370 mv_is_value_missing.
1371 (mv_is_num_user_missing): Removed. Changed callers to use
1373 (mv_is_str_user_missing): Removed. Changed callers to use
1375 (mv_is_value_system_missing): Removed. Changed callers to use
1376 mv_is_value_missing.
1377 (mv_set_type): Removed. Changed callers to use mv_clear.
1378 (mv_clear): New function.
1380 * variable.c (var_is_value_missing): Add new enum mv_class
1381 parameter. Update all callers.
1382 (var_is_num_missing): Ditto.
1383 (var_is_str_missing): Ditto.
1384 (var_is_value_user_missing): Removed. Changed callers to use
1385 var_is_value_missing.
1386 (var_is_num_user_missing): Removed. Changed callers to use
1388 (var_is_str_user_missing): Removed. Changed callers to use
1390 (var_is_value_system_missing): Removed. Changed callers to use
1391 var_is_value_missing.
1393 * casefilter.c (struct casefilter): Use enum mv_class in place of
1395 (casefilter_variable_missing): Adapt to new member.
1396 (casefilter_create): Change signature to take enum mv_class,
1399 Fri Dec 22 20:08:38 WST 2006 John Darrington <john@darrington.wattle.id.au>
1401 * casefile-factory.h fastfile-factory.c fastfile-factory.h: New files.
1403 * case-sink.c case-sink.h procedure.c procedure.h
1404 storage-stream.c: Now uses the factory.
1406 Sat Dec 16 22:05:18 2006 Ben Pfaff <blp@gnu.org>
1408 Make it possible to pull cases from the active file with a
1409 function call, instead of requiring indirection through a callback
1412 * case-source.h (struct case_source_class): Change ->read function
1413 to return a single case, instead of calling a callback function
1414 for each case. Change ->destroy function to return an error
1417 * case-source.c (free_case_source): Pass along the value returned
1418 by the case_source ->destroy function.
1420 * procedure.c (struct write_case_data): Removed.
1421 (struct dataset): Added some members to track procedure state.
1422 (procedure): Optimize the trivial case at this level.
1423 (internal_procedure): Re-implement in terms of proc_open,
1424 proc_read, proc_close.
1425 (proc_open) New function.
1426 (proc_read) New function.
1427 (proc_close) New function.
1428 (write_case) Moved into proc_read.
1429 (close_active_file) Moved closing of data source into proc_close.
1431 * storage-source.c: Rewrote to conform with modified
1432 case_source_class interface.
1434 * transformations.c (trns_chain_execute): Added argument to allow
1435 starting execution from an arbitrary transformation. Updated
1438 * transformations.h (enum TRNS_NEXT_CASE) Renamed TRNS_END_CASE.
1440 Sat Dec 16 14:09:25 2006 Ben Pfaff <blp@gnu.org>
1442 * sys-file-reader.c (read_display_parameters): Don't assume that
1443 MEASURE_* and ALIGN_* have the same values found in system files.
1445 * sys-file-writer.c (write_variable_display_parameters): Ditto.
1447 * variable.h: Change MEASURE_NOMINAL, MEASURE_ORDINAL,
1448 MEASURE_SCALE to be 0-based instead of 1-based. This also fixes
1449 the value of n_MEASURES, which was off by 1 (at least from my
1452 Sat Dec 16 12:17:34 WST 2006 John Darrington <john@darrington.wattle.id.au>
1454 * dictionary.c dictionary.h vardict.h variable.c: Added optional
1455 callbacks which are invoked when the dictionary or its
1456 variables are changed.
1458 * missing-values.c missing-values.h value-labels.c: Tidied up
1459 consistency checks, and made some of them return false
1460 instead of assert-failing.
1462 Wed Dec 13 19:30:11 2006 Ben Pfaff <blp@gnu.org>
1464 * calendar.c (calendar_days_in_month): New function.
1466 Mon Dec 11 07:53:39 2006 Ben Pfaff <blp@gnu.org>
1468 * value-labels.c (hash_int_val_lab): Only hash as many bytes as
1469 the value label's width.
1471 Sun Dec 10 14:21:29 2006 Ben Pfaff <blp@gnu.org>
1473 * sfm-private.h: Move contents into sys-file-writer.c, which is
1474 the only remaining user. Removed Borland C++-specific directives.
1476 * sys-file-reader.c: Clean up and rewrite entire file. The
1477 rewritten version is simpler and better abstracted, and should be
1478 easier to maintain and extend. It avoids using structures to read
1479 file data, which is prone to padding variations among compilers.
1480 It should also handle non-IEEE 754 system files, although I
1481 haven't been able to find any. It has been tested against many
1482 .sav files obtained from the Web and found to produce the same
1483 results as the earlier version of the code, or in some cases
1484 improved results. It is more tolerant of format variations found
1487 * sys-file-reader.h (struct sfm_read_info): Removed `big_endian'
1488 member, putting an enum integer_format in its place. New member
1489 `float_format'. Changed `compressed' member to type bool.
1491 Sun Dec 10 13:48:53 2006 Ben Pfaff <blp@gnu.org>
1493 * dictionary.c (dict_delete_consecutive_vars): New function.
1495 Sat Dec 9 20:08:25 2006 Ben Pfaff <blp@gnu.org>
1497 * file-name.c (fn_search_path): Remove prefix arg that was unused
1498 by any caller. Updated all callers.
1500 Sat Dec 9 20:04:22 2006 Ben Pfaff <blp@gnu.org>
1502 * format.c (fmt_dollar_template): Use user's decimal point
1503 character. Add assertion.
1505 Sat Dec 9 20:02:25 2006 Ben Pfaff <blp@gnu.org>
1507 * format.c (fmt_dollar_template): New function, based on
1508 dollar_format_template from var-type-dialog.c.
1510 Sat Dec 9 18:05:59 2006 Ben Pfaff <blp@gnu.org>
1512 * data-out.c (output_scientific): Fix bad assumption that "buf" is
1515 Sat Dec 9 17:23:23 2006 Ben Pfaff <blp@gnu.org>
1517 Finish converting struct variable to an opaque type. In this
1518 phase, we add remaining setter and getter functions, convert the
1519 remaining PSPP code to use them, and do a bunch of cleanup. The
1520 resulting changes are pervasive but mostly trivial, and only the
1521 notable changes are logged.
1523 * automake.mk (src_data_libdata_a_SOURCES): Add the new source
1526 * case.c (case_data): Renamed case_data_idx.
1527 (case_num): Renamed case_num_idx.
1528 (case_str): Renamed case_str_idx.
1529 (case_data_rw): Renamed case_data_rw_idx.
1531 * case.h (case_data): New function with old name and an interface
1532 that takes a variable instead of an index, which is easier to
1533 use. Updated all callers to use the new interface, or to use the
1534 new *_idx function (see above).
1537 (case_data_rw): Ditto.
1539 * category.c (cat_stored_values_destroy): Changed interface to
1540 take a struct cat_vals * instead of a struct variable *.
1542 * dictionary.c (dict_clone): Use new vector_clone function.
1543 (dict_clear) Use new var_destroy function.
1544 (add_var) New function.
1545 (dict_create_var) Rewrite in terms of dict_create_var_assert.
1546 (dict_create_var_assert) Rewrite in terms of add_var.
1547 (dict_clone_var) Rewrite in terms of dict_clone_var_assert.
1548 (dict_clone_var_assert) Rewrite in terms of var_clone, add_var.
1549 (dict_lookup_var) Use new var_create, var_destroy functions.
1550 (dict_contains_var) Rewrite in terms of new vardict functionality.
1551 (set_var_dict_index) New function.
1552 (set_var_case_index) New function.
1553 (reindex_vars) New function.
1554 (dict_delete_var) Rewrite in terms of new vardict functionality.
1555 (dict_reorder_var) Ditto.
1556 (dict_reorder_vars) Ditto.
1557 (rename_var) New function.
1558 (dict_rename_var) Use rename_var.
1559 (dict_rename_vars) Use pool to simplify code. Use rename_var.
1560 (dict_get_compacted_idx_to_fv) Rename
1561 dict_get_compacted_dict_index_to_case_index, update callers.
1562 (dict_create_vector) Use new vector_create function.
1563 (dict_clear_vectors) Use new vector_destroy function.
1564 (set_var_short_name_suffix) Move here from variable.c, renamed
1565 from var_set_short_name_suffix, make static, update caller.
1567 * sys-file-private.c: New file.
1568 (sfm_width_to_bytes) Moved here from variable.c, renamed from
1569 width_to_bytes, update callers.
1571 * sys-file-private.h: New file. Later it will supplant
1572 sfm-private.h; for now it supplements it.
1573 (macro MIN_VERY_LONG_STRING) New macro.
1574 (macro EFFECTIVE_LONG_STRING_LENGTH) New macro, from value.h.
1576 * sys-file-reader.c: Use MIN_VERY_LONG_STRING - 1 where
1577 MAX_LONG_STRING was used before.
1579 * sys-file-writer.c: Ditto.
1581 * value-labels.c: Change the paradigm here to be that a null
1582 pointer is OK for a struct val_labs * in most cases; it just
1583 represents an empty set of value labels.
1584 (val_labs_copy) A copy of a null set is a null set.
1585 (val_labs_count) A null set has 0 labels.
1586 (val_labs_replace) Change return type to void. Rewrite for
1588 (val_labs_find) A null set does not contain the value.
1589 (value_to_string) Moved to variable.c, renamed var_get_value_name,
1590 transposed argument order, updated all callers.
1592 * value.c: New file.
1593 (value_dup) Moved here from variable.c.
1594 (compare_values) Ditto.
1597 * value.h: (macro MAX_SHORT_STRING) Rewrote for simplicity.
1598 (macro MAX_LONG_STRING) Removed, because it was only interesting
1599 for system files, not for general code.
1600 (macro MAX_VERY_LONG_STRING) Ditto.
1601 (macro EFFECTIVE_LONG_STRING_LENGTH) Moved to sys-file-private.h.
1602 (macro MAX_ELEMS_PER_VALUE) Removed, as it was unused.
1604 * vardict.h: New file, for an interface between variables and
1607 * variable.c: A lot of functions were moved around, for better
1609 (struct variable) Move definition here, from variable.h.
1610 (var_type_adj) Removed--makes i18n hard.
1611 (var_type_noun) Ditto.
1612 (var_create) New function.
1613 (var_clone) New function.
1614 (var_destroy) New function.
1615 (var_set_name) Assert that variable is not in a dictionary.
1616 (compare_var_names) Rename compare_vars_by_name and fix a couple
1617 of callers who thought the args were strings.
1618 (hash_var_name) Rename hash_var_by_name.
1619 (compare_var_ptr_names) Rename compare_var_ptrs_by_name.
1620 (hash_var_ptr_name) Rename hash_var_ptr_by_name.
1621 (var_is_very_long_string) Removed, because it was only interesting
1622 to system file code.
1623 (var_set_missing_values) Allow the argument to be the wrong width,
1624 as long as we can resize it. Simplify callers who were doing the
1625 resizing themselves.
1626 (var_get_value_labels) New function.
1627 (var_has_value_labels) New function.
1628 (var_set_value_labels) New function.
1629 (alloc_value_labels) New function.
1630 (var_add_value_label) New function.
1631 (var_replace_value_label) New function.
1632 (var_clear_value_labels) New function.
1633 (var_lookup_value_label) New function.
1634 (var_get_value_name) Moved here from variable.c, renamed from
1635 var_get_value_name, transposed argument order, updated all
1637 (var_to_string) Moved here, from variable-label.c.
1638 (var_set_leave) New function.
1639 (var_get_leave) New function.
1640 (var_must_leave) New function.
1641 (var_set_short_name_suffix) Moved to dictionary.c, renamed
1642 set_var_short_name_suffix.
1643 (var_get_dict_index) New function.
1644 (var_get_case_index) New function.
1645 (var_get_obs_vals) New function.
1646 (var_set_obs_vals) New function.
1647 (var_has_obs_vals) New function.
1648 (var_get_vardict) New function.
1649 (var_set_vardict) New function.
1650 (var_has_vardict) New function.
1651 (var_clear_vardict) New function.
1652 (value_dup) Moved to value.c.
1653 (compare_values) Ditto.
1656 * variable.h: (enum NUMERIC) Rename VAR_NUMERIC, update all users.
1657 (enum ALPHA) Rename VAR_STRING, update all users.
1659 * vector.c: New file.
1660 (struct vector) Moved here, from variable.h.
1661 (check_widths) New function.
1662 (vector_create) New function.
1663 (vector_clone) New function.
1664 (vector_destroy) New function.
1665 (vector_get_name) New function.
1666 (vector_get_var) New function.
1667 (vector_get_var_cnt) New function.
1668 (compare_vector_ptrs_by_name) New function.
1670 * vector.h: New file.
1672 Sun Dec 10 11:32:56 WST 2006 John Darrington <john@darrington.wattle.id.au>
1674 * casefilter.c (casefilter_variable_missing): Avoided comparision of
1675 string variables to SYSMIS. Thanks to Ben Pfaff for reporting this
1678 Sat Dec 9 07:18:03 WST 2006 John Darrington <john@darrington.wattle.id.au>
1680 * value-labels.c (destroy_atoms): New function.
1681 * value-labels.c (atom_create): Call destroy_atoms in atexit handler.
1683 Thu Dec 7 17:38:26 2006 Ben Pfaff <blp@gnu.org>
1685 Thanks to Jason Stover for pointing out this problem.
1687 * data-out.c (output_number): Use gsl_finite from GSL, which is
1688 portable, instead of isfinite, which is not.
1691 Thu Dec 7 15:22:38 WST 2006 John Darrington <john@darrington.wattle.id.au>
1693 * variable.c variable.h (value_dup): New function.
1695 Mon Dec 4 22:20:17 2006 Ben Pfaff <blp@gnu.org>
1697 Start converting struct variable to an opaque type. In this
1698 phase, we add a bunch of setter and getter functions and convert
1699 most of the PSPP code to use them. The resulting changes are
1700 pervasive but mostly trivial, and only the notable changes are
1703 * format.c (fmt_equal): New function.
1705 * variable.c (var_type_is_valid): New function.
1706 (measure_is_valid) Moved here, from format.c.
1707 (alignment_is_valid) Moved here, from format.c.
1708 (var_get_name) New function.
1709 (var_set_name) New function.
1710 (width_to_type) New function.
1711 (var_get_type) New function.
1712 (var_get_width) New function.
1713 (var_set_width) New function.
1714 (var_is_numeric) New function.
1715 (var_is_alpha) New function.
1716 (var_is_short_string) New function.
1717 (var_is_long_string) New function.
1718 (var_is_very_long_string) New function.
1719 (var_get_missing_values) New function.
1720 (var_set_missing_values) New function.
1721 (var_clear_missing_values) New function.
1722 (var_has_missing_values) New function.
1723 (var_is_value_missing) New function.
1724 (var_is_num_missing) New function.
1725 (var_is_str_missing) New function.
1726 (var_is_value_user_missing) New function.
1727 (var_is_num_user_missing) New function.
1728 (var_is_str_user_missing) New function.
1729 (var_is_value_system_missing) New function.
1730 (var_get_print_format) New function.
1731 (var_set_print_format) New function.
1732 (var_get_write_format) New function.
1733 (var_set_write_format) New function.
1734 (var_set_both_formats) New function.
1735 (var_get_label) New function.
1736 (var_set_label) New function.
1737 (var_clear_label) New function.
1738 (var_has_label) New function.
1739 (var_get_measure) New function.
1740 (var_set_measure) New function.
1741 (var_get_display_width) New function.
1742 (var_set_display_width) New function.
1743 (var_get_alignment) New function.
1744 (var_set_alignment) New function.
1745 (var_get_value_cnt) New function.
1746 (var_get_leave) New function.
1747 (var_get_short_name) New function.
1749 * variable.h: (struct variable) Removed "type" and "nv" members;
1750 they are now computed from "width" where needed.
1752 Mon Dec 4 21:38:40 2006 Ben Pfaff <blp@gnu.org>
1754 * missing-values.c (mv_resize): Don't write beyond end of the
1755 allocated buffer when resizing a long string.
1757 Sat Dec 2 16:28:32 2006 Ben Pfaff <blp@gnu.org>
1759 Clean up identifier code: don't require identifier enumerations to
1760 be in a particular order; make better use of string library;
1761 expose less of the internals.
1763 * identifier.c: (lex_skip_identifier) Rename lex_id_get_length,
1764 change interface. Updated all callers.
1765 (lex_id_match) Change interface to use struct substring, update
1767 (lex_id_match_len) Removed. Update callers to use lex_id_match.
1768 (global array keywords[]) Make static, change form. Update all
1769 users to use lex_id_name instead.
1770 (lex_is_keyword) New function.
1771 (lex_id_to_token) Change interface to use struct substring, update
1773 (lex_id_name) New function.
1775 * identifier.h: (T_FIRST_KEYWORD) Removed. Changed users to call
1776 lex_is_keyword instead.
1777 (T_LAST_KEYWORD) Removed.
1778 (T_N_KEYWORDS) Removed.
1780 Sat Nov 18 20:46:35 2006 Ben Pfaff <blp@gnu.org>
1782 * format.c: (fmt_date_template) Distinguish characters for which a
1783 space is output and any date delimiter is allowed on input, from
1784 those for which a space is output and only a space is allowed on
1785 input. The former is represented by X, the latter by a space.
1786 Also, drop distinction between h and H, changing the former to the
1789 * data-in.c: Completely rewrite internals to conform to SPSS input
1790 formats as closely as possible.
1791 (data_in) Changed external interface by replacing the structure
1792 that was used as a single argument by a set of arguments. Updated
1794 (data_in_finite_line) Removed. Converted all callers to use plain
1796 (data_in_get_integer_format) New function.
1797 (data_in_set_integer_format) New function.
1798 (data_in_get_float_format) New function.
1799 (data_in_set_float_format) New function.
1801 * data-in.h: (enums DI_IGNORE_ERROR, DI_IMPLIED_DECIMALS) Removed.
1802 (struct data_in) Removed.
1804 * data-out.c: (output_date) Drop each component from the input as
1805 it is output, to allow us to drop the distinction between h (a
1806 count of hours) and H (the hour of day) template characters.
1807 Also, handle new X template character.
1808 (output_scientific) Follow more rational rule on when to drop
1809 fraction introduced between SPSS 13 and 15. Updated test case to
1812 Sat Nov 11 11:41:26 2006 Ben Pfaff <blp@gnu.org>
1814 Fix buffer overflow reported by John Darrington.
1816 * data-out.c (output_bcd_integer): In case of SYSMIS, etc.,
1817 realize that DIGITS is a count of nibbles, not of bytes.
1819 Sat Nov 4 15:59:56 2006 Ben Pfaff <blp@gnu.org>
1821 * calendar.c (calendar_offset_to_gregorian) Also return the
1822 year-of-day. Change callers to new interface.
1824 * data-out.c: Completely rewrite internals to conform to SPSS
1825 output formats as completely as possible.
1826 (data_out) Change interface to put input parameters before output
1827 parameters, for consistency with the style I now prefer. Update
1829 (data_out_get_integer_format) New public function.
1830 (data_out_set_integer_format) New public function.
1831 (data_out_get_float_format) New public function.
1832 (data_out_set_float_format) New public function.
1834 * data-out.h: New file. Move prototype for data_out here, from
1837 * format.c: (fmt_step_width) Use equality comparison instead of
1838 bitwise and, for clarity.
1839 (fmt_is_string) Ditto.
1840 (fmt_input_to_output) Fix categories that are translated to F
1843 Sun Nov 5 08:29:34 WST 2006 John Darrington <john@darrington.wattle.id.au>
1845 * casefilter.c casefilter.h (new files), casefile.c casefile.h
1846 casefile-private.h: Added casefilter to assist commands with missing
1849 Sat Nov 4 11:47:09 2006 Ben Pfaff <blp@gnu.org>
1851 Implement SET ERRORS, SHOW ERRORS. Fixes bug #17609.
1853 * settings.c: (route_errors_to_terminal) New variable.
1854 (route_errors_to_listing) New variable.
1855 (get_error_routing_to_terminal) New function.
1856 (set_error_routing_to_terminal) New function.
1857 (get_error_routing_to_listing) New function.
1858 (set_error_routing_to_listing) New function.
1860 * settings.h: (SET_ROUTE_* enums) Removed, because unused.
1862 Tue Oct 31 19:58:27 2006 Ben Pfaff <blp@gnu.org>
1864 * format.c: Completely rewrite, to achieve better abstraction.
1865 Rewrite all references to formats in other files.
1867 * format.def: Rewrite and reorganize.
1869 * settings.c: Move everything related to custom currency formats
1870 into format.[ch], changing them in form, so as to group related
1871 code and definitions better. Changed all references to use the
1873 (static var decimal) Removed.
1874 (static var grouping) Removed.
1875 (static var cc) Removed.
1876 (get_decimal) Removed.
1877 (set_decimal) Removed.
1878 (get_grouping) Removed.
1879 (set_grouping) Removed.
1883 * settings.h: (macro CC_CNT) Removed.
1884 (macro CC_WIDTH) Removed.
1885 (struct custom_currency) Removed.
1887 Tue Oct 31 19:56:19 2006 Ben Pfaff <blp@gnu.org>
1889 * data-in.c (data_in): Use switch statement instead of table, to
1890 avoid dependence on the order of the FMT_* enums.
1892 Tue Oct 31 19:35:36 2006 Ben Pfaff <blp@gnu.org>
1894 * data-out.c: (num_to_string) Removed, because it was dead code.
1896 Tue Oct 31 18:09:24 2006 Ben Pfaff <blp@gnu.org>
1898 * data-in.c (parse_trailer): Fix error message.
1900 Sat Oct 28 11:56:50 2006 Ben Pfaff <blp@gnu.org>
1902 * format.c (fmt_is_binary): New function.
1904 Thu Oct 19 22:59:56 WST 2006 John Darrington <john@darrington.wattle.id.au>
1906 * procedure.c procedure.h: Encapsulated the static data into a single
1909 Sat Oct 14 16:56:44 2006 Ben Pfaff <blp@gnu.org>
1911 * casefile.c (casereader_read_xfer): Always initialize the case,
1912 even on an error condition.
1914 Wed Sep 27 09:37:49 WST 2006 John Darrington <john@darrington.wattle.id.au>
1916 * procedure.c (case_limit_trns_proc): Fixed buglet which rendered the
1917 entire function useless.
1919 Mon Sep 25 17:11:46 WST 2006 John Darrington <john@darrington.wattle.id.au>
1921 * casefile-private.h casefile.c casefile.h fastfile.c: Created new
1922 casereader method casereader_clone.
1924 * procedure.c pransformations.h: Introduced new type casenum_t
1926 Thu Sep 21 07:00:30 2006 Ben Pfaff <blp@gnu.org>
1928 * variable.c: (width_to_bytes) Rephrase code for clarify.
1930 Sun Jul 16 19:52:03 2006 Ben Pfaff <blp@gnu.org>
1932 * format.c: (fmt_type_from_string) New function.
1933 (fmt_to_string) Include decimals in output if the format has
1934 decimals, even if the format type does not. This way, we can
1935 accurately reproduce incorrect formats in user output.
1936 (check_common_specifier) Make the check for a bad format type an
1937 assertion, so we get bug reports if they show up. Fix message.
1938 Check for decimal places with a format type that doesn't allow
1940 (check_input_specifier) Remove check for FMT_X, which has been
1942 (check_output_specifier) Ditto.
1944 * format.def: Remove FMT_T, FMT_X, FMT_DESCEND, FMT_NEWREC.
1946 * format.h: (macro FMT_TYPE_LEN_MAX) New macro.
1947 (struct fmt_desc) Use FMT_TYPE_LEN_MAX in definition.
1948 (enum fmt_parse_flags) Removed.
1950 Mon Jul 17 18:26:21 WST 2006 John Darrington <john@darrington.wattle.id.au>
1952 * casefile.c casefile.h: Converted to an abstract base class.
1953 * casefile-private.h fastfile.c fastfile.h: New files.
1954 * automake.mk procedure.c scratch-writer.c storage-stream.c
1956 Wed Jul 12 21:02:26 2006 Ben Pfaff <blp@gnu.org>
1958 * procedure.c (internal_procedure): Create sink_case with only as
1959 many values as the compacted dictionary.
1961 Wed Jul 12 21:01:00 2006 Ben Pfaff <blp@gnu.org>
1963 Remove "debugging" code that caused plenty of false positives and
1966 * case.h (struct ccase): [DEBUGGING] Remove `this' member.
1968 * case.c: Remove all references to `this' member.
1970 Thu Jul 6 19:09:53 2006 Ben Pfaff <blp@gnu.org>
1972 Fix link error noted by Jason Stover.
1974 * storage-stream.c: Include <assert.h>.
1976 Tue Jul 4 08:47:35 2006 Ben Pfaff <blp@gnu.org>
1978 Fix bug #15766 (/KEEP subcommand on SAVE doesn't fully support
1979 ALL) and additional underlying system file issues.
1981 Thanks to John Darrington for review.
1983 First problem: var_hash points to variables not owned by the
1984 sys-file-reader, which the caller may free or modify. Use an
1985 array of sfm_vars instead, as done earlier (e.g. CVS version
1988 * sys-file-reader.c (struct sfm_reader): Remove var_hash, svars
1989 members and remove all code that references it. Add vars, var_cnt
1990 members. Remove fix_specials member, which was unused.
1991 (struct sfm_var) Remove name member, which was unused.
1992 (sfm_close_reader) Free vars member instead of var_hash.
1993 (compare_var_shortnames) Removed.
1994 (hash_var_shortname) Removed.
1995 (sfm_open_reader) Fill out vars array.
1996 (compare_var_index) Removed.
1997 (sfm_read_case) Use vars instead of var_hash.
1999 Second problem: we're confused about when we actually have very
2000 long strings, causing us to choose incorrectly between slow path
2001 and fast path in sfm_read_case.
2003 * sys-file-reader.c: (sfm_open_reader) Only mark has_vls if we
2004 have very long strings, not when we have long variable names,
2005 which is an unrelated feature.
2007 Tue Jun 27 12:06:49 2006 Ben Pfaff <blp@gnu.org>
2009 * variable.h: Move var_set and variable parsing declarations to
2010 new header, src/language/lexer/variable-parser.h. Modified lots
2011 of files to include the new header.
2013 Sun Jun 25 22:39:32 2006 Ben Pfaff <blp@gnu.org>
2015 * value-labels.c (value_to_string): When there's no value label,
2016 format the variable according to its print format, instead of
2017 always effectively using A or F format.
2019 Mon Jun 19 18:05:42 WST 2006 John Darrington <john@darrington.wattle.id.au>
2021 * casefile.c (casefile_get_random_reader): Nasty hack to get around
2024 * format.c: Removed tortological assertion.
2026 Fri Jun 9 12:20:09 2006 Ben Pfaff <blp@gnu.org>
2028 Reform string library.
2030 * file-name.c (fn_interp_vars): Change interface to take a
2031 substring as input. Updated all users.
2033 Fri Jun 9 12:11:24 2006 Ben Pfaff <blp@gnu.org>
2035 * format.c (measure_is_valid): Really return false when m >=
2038 Tue Jun 6 18:46:26 2006 Ben Pfaff <blp@gnu.org>
2040 Implement random access to casefiles, for use in GUI.
2042 * casefile.c: (struct casereader) Add `random', `file_ofs',
2043 `buffer_ofs' members.
2044 (casefile_get_random_reader) New function.
2045 (read_open_file) Break part into new function
2046 seek_and_fill_buffer().
2047 (fill_buffer) Update buffer_ofs, file_ofs.
2048 (casereader_seek) New function.
2050 Tue May 30 19:52:33 WST 2006 John Darrington <john@darrington.wattle.id.au>
2052 * settings.c: Added call to i18n{done, init}.
2054 Tue May 9 21:09:17 2006 Ben Pfaff <blp@gnu.org>
2056 * procedure.h: Add WARN_UNUSED_RESULT to procedure function
2059 Tue May 9 21:08:05 2006 Ben Pfaff <blp@gnu.org>
2061 * casefile.c: Convert many uses of `int' to `bool'.
2063 Sat May 6 22:49:43 2006 Ben Pfaff <blp@gnu.org>
2065 * transformations.c (trns_chain_destroy): Destroy chain's trns
2066 member, to fix memory leak.
2068 Sat May 6 22:48:30 2006 Ben Pfaff <blp@gnu.org>
2070 * storage-stream.c (storage_source_decapsulate): Destroy case
2071 source to fix memory leak.
2073 Sat May 6 22:46:47 2006 Ben Pfaff <blp@gnu.org>
2075 * scratch-reader.c (scratch_reader_read_case): Copy into existing
2076 case passed as argument instead of initializing the argument as a
2077 case. Fixes memory leak that showed up in
2078 tests/command/aggregate.sh with scratch files.
2080 Sat May 6 22:45:55 2006 Ben Pfaff <blp@gnu.org>
2082 * procedure.c (proc_done): Destroy default_dict, to fix memory
2085 Sat May 6 22:44:44 2006 Ben Pfaff <blp@gnu.org>
2087 Simplify procedure_with_splits().
2089 * procedure.c (struct split_aux_data): Removed case_count member.
2090 (procedure_with_splits) Don't initialize case_count.
2091 (split_procedure_case_func) Check whether prev_case is null
2092 instead of case_count.
2093 (split_procedure_end_func) Ditto.
2095 Sat May 6 22:42:23 2006 Ben Pfaff <blp@gnu.org>
2097 * case.c (case_move): Do nothing if dst and src are the same
2099 (case_try_create) Merge two similar cases.
2100 (case_copy) Unshare only if data must be actually copied.
2102 Sun May 7 10:04:06 WST 2006 John Darrington <john@darrington.wattle.id.au>
2104 * data-in.c data-out.c dictionary.c sys-file-reader.c
2105 sys-file-writer.c variable.c variable.h: Reworked very long string
2106 support for better encapsulation.
2108 Sat May 6 19:02:00 2006 Ben Pfaff <blp@gnu.org>
2110 * value-labels.c (val_labs_can_set_width): New function.
2111 (val_labs_set_width) Clear labels if increasing width to long
2113 (val_labs_destroy) Remove unneeded test for null.
2115 Sat May 6 16:14:08 2006 Ben Pfaff <blp@gnu.org>
2117 * value-labels.h: Remove unneeded dependency on variable.h.
2119 Sat May 6 15:58:36 2006 Ben Pfaff <blp@gnu.org>
2121 Get rid of `char *c' member in union value, for cleanliness.
2123 * value.h: (union value) Remove `c' member.
2125 Sat May 6 15:36:59 2006 Ben Pfaff <blp@gnu.org>
2127 Make dictionary compacting functions a little more general.
2129 * sys-file-writer.c (sfm_open_writer): Use
2130 dict_compacting_would_change().
2131 (does_dict_need_translation) Removed.
2133 Sat May 6 15:35:42 2006 Ben Pfaff <blp@gnu.org>
2135 Make dictionary compacting functions a little more general.
2137 * dictionary.c (dict_needs_compaction): Rename
2138 dict_compacting_would_shrink(). Update all callers.
2139 (dict_compacting_would_change) New function.
2141 Sat May 6 14:25:49 2006 Ben Pfaff <blp@gnu.org>
2143 * sys-file-writer.c: (does_dict_need_translation) Fix bug:
2144 inverted return value (!).
2146 Sat May 6 13:37:52 2006 Ben Pfaff <blp@gnu.org>
2148 Continue reforming procedure execution.
2150 * procedure.c: Search and replace "vfm" by "proc". Notably:
2151 (static var vfm_source) Rename proc_source. Update all
2153 (static var vfm_sink) Rename proc_sink. Update all references.
2155 Sat May 6 12:38:55 2006 Ben Pfaff <blp@gnu.org>
2157 Continue reforming procedure execution. In this phase, remove
2158 PROCESS IF, which was deprecated anyway and can be easily
2159 simulated with TEMPORARY followed by SELECT IF.
2161 * procedure.c: (open_active_file) Don't call
2162 add_process_if_trns().
2163 (discard_variables) Get rid of redundant call to
2164 proc_cancel_all_transformations().
2165 (add_process_if_trns) Removed.
2166 (process_if_trns_proc) Removed.
2167 (process_if_trns_free) Removed.
2169 Sat May 6 10:58:05 2006 Ben Pfaff <blp@gnu.org>
2171 Continue reforming procedure execution. In this phase, add
2172 `const' to the case passed to procedure()'s callback.
2174 Updated all users of procedure() as well.
2176 * procedure.c: (struct write_case_data) Add "const" to ccase
2177 parameter for case_func member.
2178 (procedure) Add "const" to ccase parameter for proc_func
2180 (multipass_case_func) Make ccase parameter const.
2181 (internal_procedure) Add "const" to ccase parameter for case_func
2183 (split_procedure_case_func) Make ccase parameter const.
2184 (multipass_split_case_func) Make ccase parameter const.
2186 Sat May 6 10:30:33 2006 Ben Pfaff <blp@gnu.org>
2188 Continue reforming procedure execution. In this phase, get rid of
2189 the output code for SPLIT FILE groups in procedure.c, which really
2190 shouldn't be doing any output. Move it into the individual
2191 procedures instead. This also adds some flexibility.
2193 Updated many users of procedure_with_splits() and
2194 multipass_procedure_with_splits() to call
2195 output_split_file_values() and to deal with increased use of
2198 * procedure.c: (struct split_aux_data) Add "const struct ccase *"
2199 parameter to begin_func member.
2200 (procedure_with_splits) Add "const struct ccase *" parameter to
2201 begin_func parameter. Make ccase parameter const in proc_func
2203 (split_procedure_case_func) Don't dump split file group. Pass
2205 (dump_splits) Moved to language/dictionary/split-file.c as
2206 output_split_file_values().
2207 (struct multipass_split_aux_data) Add "const struct ccase *"
2208 parameter to split_func member.
2209 (multipass_procedure_with_splits) Add "const struct ccase *"
2210 parameter to split_func parameter.
2211 (multipass_split_case_func) Save new SPLIT FILE case before
2213 (multipass_split_output) Pass saved SPLIT FILE case to split_func.
2215 Fri May 5 22:48:50 2006 Ben Pfaff <blp@gnu.org>
2217 Continue reforming procedure execution. Change
2218 internal_procedure() so that it calls open_active_file() and
2219 close_active_file(), which isolates most of the actual procedure
2222 * procedure.c: (struct write_case_data) Rename `proc_func' member
2223 to `case_func' and update all references.
2224 (procedure) Rewrite as one-line wrapper around
2225 internal_procedure().
2226 (struct multipass_aux_data) New.
2227 (multipass_callback) Renamed multipass_case_func(). Use struct
2228 multipass_aux_data as auxiliary data.
2229 (multipass_end_func) New function.
2230 (multipass_procedure) Rewrite as wrapper for internal_procedure()
2231 that uses multipass_case_func, multipass_end_func.
2232 (internal_procedure) Add `end_func' argument. Move optimization
2233 of trivial case in here. Move call to open_active_file() and
2234 close_active_file() in here. Now assert that vfm_source is
2236 (procedure_with_splits_callback) Rename
2237 split_procedure_case_func().
2238 (split_procedure_end_func) New function.
2239 (multipass_split_callback) Rename multipass_split_case_func.
2240 (multipass_split_end_func) New function.
2241 (discard_variables) No need to test for nonnull vfm_source.
2243 Fri May 5 21:34:02 2006 Ben Pfaff <blp@gnu.org>
2245 Continue reforming procedure execution. Get rid of unused member.
2247 * procedure.c: (struct write_case_data) Remove `cases_analyzed'
2249 (write_case) Don't increment cases_analyzed.
2251 Thu May 4 21:50:11 2006 Ben Pfaff <blp@gnu.org>
2253 Continue reforming procedure execution. In this phase, move
2254 procedure.c and procedure.h from src to src/data. Update
2255 makefiles and #includes accordingly.
2257 * procedure.c: Moved here from src/.
2259 * procedure.h: Moved here from src/.
2261 Wed May 3 22:42:12 2006 Ben Pfaff <blp@gnu.org>
2263 Continue reforming procedure execution. In this phase, get rid of
2264 many global variables, consolidating procedure execution in
2265 procedure.c. Encapsulate transformations in new "struct
2266 trns_chain". Also, change implementation of N OF CASES, FILTER,
2267 and PROCESS IF from special cases to transformations.
2269 * automake.mk: (src_data_libdata_a_SOURCES) Add transformations.c,
2272 * dictionary.c: (global variable default_dict) Move to
2275 * variable.h: (TRNS_*) Move to transformations.h.
2276 (struct transformation) Move to transformations.c.
2278 Thu May 4 13:47:06 WST 2006 John Darrington <john@darrington.wattle.id.au>
2280 * sys-file-reader.c: Fixed invalid read problems.
2282 Tue May 2 15:57:10 2006 Ben Pfaff <blp@gnu.org>
2284 * storage-stream.c: Add missing function comments.
2286 Tue May 2 15:50:21 2006 Ben Pfaff <blp@gnu.org>
2288 Continue reforming procedure execution. In this phase, add some
2289 new, needed functionality to storage-stream.
2291 * storage-stream.c: (storage_source_decapsulate) New function.
2293 Tue May 2 15:43:36 2006 Ben Pfaff <blp@gnu.org>
2295 * variable.c (width_to_bytes): Declarations must precede
2296 statements for C90 compliance.
2298 Tue May 2 10:42:05 WST 2006 John Darrington <john@darrington.wattle.id.au>
2300 * data-out.c, data-in.c, variable.c, variable.h: New functions
2301 copy_mangle and copy_demangle for reading/writing cases; emulates the
2302 way SPSS deals with strings > 255 bytes.
2304 * sys-file-reader.c sys-file-writer.c: Added support for Record 7,
2305 subtype 14 needed for strings longer than 255 bytes.
2307 * dictionary.c, format.def, value.c : Updated to use MAX_STRING
2308 instead of literal values. Also fixed some constness issues.
2310 * format.h: Constness
2312 * sfm-private.h: Renamed the case_size identifier, since I discovered
2313 that SPSS's respect for this variable is very nominal.
2315 Mon May 1 15:45:42 2006 Ben Pfaff <blp@gnu.org>
2317 Change case limit type from int to size_t.
2319 * dictionary.c: (struct dictionary) Change type of case_limit
2321 (dict_get_case_limit) Change return type.
2322 (dict_set_case_limit) Change parameter type.
2324 Wed Apr 26 20:01:19 2006 Ben Pfaff <blp@gnu.org>
2326 * variable.h: (struct variable) Rename `reinit' member as `leave'
2327 and invert sense. Fix up all references.
2329 Wed Apr 26 19:39:28 2006 Ben Pfaff <blp@gnu.org>
2331 Continue reforming procedure execution. In this phase, break
2332 procedure.c into multiple files.
2334 * automake.mk: (src_data_libdata_a_SOURCES) Add all the new files.
2336 * case-sink.c: New file.
2338 * case-sink.h: New file.
2340 * case-source.c: New file.
2342 * case-source.h: New file.
2344 * storage-stream.c: New file.
2346 * storage-stream.h: New file.
2348 Wed Apr 26 14:55:19 2006 Ben Pfaff <blp@gnu.org>
2350 * variable.h: (struct variable) Remove `init' member and all
2351 references to it from other files. It was initialized in several
2352 places, but nothing really ever used it for anything worthwhile.
2353 Thanks to Jason Stover for pointing out how confusing this
2356 Sun Apr 23 22:04:45 2006 Ben Pfaff <blp@gnu.org>
2358 Continue reforming error message support. In this phase, get rid
2359 of message "titles" and put the message text in `struct error'.
2360 Now `struct error' encapsulates a message more properly.
2362 * casefile.c: (io_error) Use err_msg() instead of err_vmsg().
2363 Format message ourselves.
2365 * data-in.c: (vdls_error) Ditto.
2367 * por-file-reader.c: (error) Ditto.
2369 * sys-file-reader.c: (corrupt_msg) Ditto.
2371 Sun Apr 16 18:49:51 2006 Ben Pfaff <blp@gnu.org>
2373 GNU standards require "file name" instead of "filename" in
2374 documentation. It's nice for our code to follow the convention
2377 * casefile.c: (struct casefile) Rename `filename' member to
2378 `file_name'. Updated all references.
2380 * file-name.c: [!unix] (struct file_identity) Rename
2381 normalized_filename member to normalized_file_name. Updated all
2384 Sun Apr 16 18:35:33 2006 Ben Pfaff <blp@gnu.org>
2386 We don't really support anything but Unix-like environments well,
2387 so we might as well de-obfuscate by writing directory and path
2388 separators explicitly.
2390 * file-name.h: (macro DIR_SEPARATOR) Removed. Changed all usages
2392 (macro PATH_SEPARATOR) Removed. Changed all usages to just ':'.
2393 (macro DIR_SEPARATOR_STRING) Removed. Changed all usages to just
2395 (macro PATH_SEPARATOR_STRING) Removed. Changed all usages to just
2398 Sun Apr 16 18:28:35 2006 Ben Pfaff <blp@gnu.org>
2400 GNU standards require "file name" instead of "filename" in
2401 documentation. It's nice for our code to follow the convention
2404 * filename.c: Rename to file-name.c.
2406 * filename.h: Rename to file-name.h. Update all inclusions.
2407 Update header guards.
2409 * automake.mk: Update file names.
2411 Sun Apr 16 16:42:47 2006 Ben Pfaff <blp@gnu.org>
2413 * filename.c: (fn_dirname) Renamed fn_dir_name(), all references
2415 (fn_basename) Removed (dead code).
2416 (fn_absolute_p) Renamed fn_is_absolute(), all references updated.
2417 (fn_special_p) Renamed fn_is_special(), all references updated.
2418 (fn_exists_p) Renamed fn_exists(), all references updated.
2420 Sun Apr 16 16:33:58 2006 Ben Pfaff <blp@gnu.org>
2422 * filename.c: (fn_tilde_expand) Rewrite for cleaner code.
2423 Also, now it only tilde-expands file names, not paths.
2424 (fn_search_path) Tilde-expand one directory at a time.
2426 Sun Apr 16 16:28:06 2006 Ben Pfaff <blp@gnu.org>
2428 * filename.c: (fn_search_path) rewrite for cleaner code. Also,
2429 get rid of non-Unixlike version of the code, which has probably
2431 (fn_prepend_dir) Removed (dead code).
2433 * filename.h: (macro DIR_SEPARATOR_STRING) New.
2434 (macro PATH_SEPARATOR_STRING) New.
2435 Sun Apr 16 16:05:28 2006 Ben Pfaff <blp@gnu.org>
2437 Continue reforming error message support. In this phase, we get
2438 rid of VM() and the other msg() support for "verbosity", replacing
2439 it by a new function verbose_msg().
2441 * filename.c: (fn_search_path) Use verbose_msg() instead of
2444 Sat Apr 15 19:53:19 2006 Ben Pfaff <blp@gnu.org>
2446 * sfm-private.h: Get rid of #defines after #error, which makes no
2449 Sat Apr 15 19:48:57 2006 Ben Pfaff <blp@gnu.org>
2451 Get rid of our own int32 type in favor of the standard int32_t
2454 * sfm-private.h: (int32 macro) Don't define this anymore. Do
2457 * sys-file-reader.c: Use int32_t instead of int32 throughout.
2459 * sys-file-writer.c: Use int32_t instead of int32 throughout.
2461 Sat Apr 15 19:36:47 2006 Ben Pfaff <blp@gnu.org>
2463 Remove ill-considered file routines that are no longer used.
2465 * filename.c: (fn_open_ext) Removed.
2466 (fn_close_ext) Removed.
2468 * filename.h: (struct file_ext) Removed.
2470 Mon Apr 3 13:22:39 2006 Ben Pfaff <blp@gnu.org>
2472 * variable.c (var_is_valid_name): Move declarations before code
2475 Tue Apr 4 15:28:40 WST 2006 John Darrington <john@darrington.wattle.id.au>
2477 * filename.ch (fn_interp_vars): Fixed small buglet.
2479 Tue Mar 28 13:47:16 WST 2006 John Darrington <john@darrington.wattle.id.au>
2481 * filename.[ch] (fn_interp_vars): Changed the signature and semantics
2482 so as to modify the string inline. Thus makeing it easier to
2483 destroy the results when no longer needed.
2485 2006-03-25 Jason Stover <jhs@math.gcsu.edu>
2487 * category.c (cat_stored_values_destroy): Fixed memory leak.
2489 Fri Mar 24 18:15:41 2006 Ben Pfaff <blp@gnu.org>
2491 Add some missing frees. Thanks to John Darrington for reporting
2494 * any-writer.c (any_writer_close): Free writer.
2496 * any-reader.c (any_reader_close): Free reader.
2498 Mon Mar 20 16:33:53 2006 Ben Pfaff <blp@gnu.org>
2500 * por-file-reader.c: (error) Mark as NO_RETURN.
2502 Sat Mar 11 15:06:07 WST 2006 John Darrington <john@darrington.wattle.id.au>
2504 * settings.c: Changed default value of scompress to true.
2506 Sat Mar 4 13:22:51 2006 Ben Pfaff <blp@gnu.org>
2508 * sfm-private.h: Include variable.h, to get SHORT_NAME_LEN.
2510 * value.h: Remove check on MAX_SHORT_STRING, which I don't think
2513 * variable.h: Move definition of SHORT_NAME_LEN, LONG_NAME_LEN
2514 here from pref.h.orig.
2516 Sat Mar 4 12:50:48 WST 2006 John Darrington <john@darrington.wattle.id.au>
2518 * sys-file-reader.c: Fixed bug reading compressed files.
2520 Thu Mar 2 08:40:33 WST 2006 John Darrington <john@darrington.wattle.id.au>
2522 * Numerous renames. See src/ChangeLog for details.
2524 * Moved files from src directory