pspp
11 years agoRename sheets to n_sheets
John Darrington [Thu, 14 Feb 2013 19:36:31 +0000 (20:36 +0100)]
Rename sheets to n_sheets

11 years agoEnable the syntax generation
John Darrington [Thu, 14 Feb 2013 18:37:56 +0000 (19:37 +0100)]
Enable the syntax generation

11 years agoMake the data visible
John Darrington [Thu, 14 Feb 2013 18:02:49 +0000 (19:02 +0100)]
Make the data visible

11 years agoUpdated the test widget thing a bit
John Darrington [Thu, 14 Feb 2013 13:35:45 +0000 (14:35 +0100)]
Updated the test widget thing a bit

11 years agoUse the gnumeric_reopen function
John Darrington [Thu, 14 Feb 2013 10:23:13 +0000 (11:23 +0100)]
Use the gnumeric_reopen function

11 years agoAdded a reopen capability
John Darrington [Thu, 14 Feb 2013 09:51:46 +0000 (10:51 +0100)]
Added a reopen capability

11 years agoDont show range if it is null
John Darrington [Tue, 29 Jan 2013 19:04:59 +0000 (20:04 +0100)]
Dont show range if it is null

11 years agoAdded a proper function to convert spreadsheet column refs
John Darrington [Tue, 29 Jan 2013 18:26:23 +0000 (19:26 +0100)]
Added a proper function to convert spreadsheet column refs

11 years agoAdded a callback to set the range entry on change
John Darrington [Sun, 27 Jan 2013 19:41:07 +0000 (20:41 +0100)]
Added a callback to set the range entry on change

11 years agoMake both work at the same time
John Darrington [Sun, 27 Jan 2013 19:09:21 +0000 (20:09 +0100)]
Make both work at the same time

11 years agoFix memory leak
John Darrington [Sun, 27 Jan 2013 18:36:18 +0000 (19:36 +0100)]
Fix memory leak

11 years agoAdded the range column to the model
John Darrington [Sun, 27 Jan 2013 18:17:41 +0000 (19:17 +0100)]
Added the range column to the model

11 years agoAdded a maxcol state
John Darrington [Sun, 27 Jan 2013 10:38:18 +0000 (11:38 +0100)]
Added a maxcol state

11 years agoTidied up a bit more. Eliminated some warnings
John Darrington [Sun, 27 Jan 2013 09:03:35 +0000 (10:03 +0100)]
Tidied up a bit more.  Eliminated some warnings

11 years agoTidied up a bit
John Darrington [Sun, 27 Jan 2013 08:50:03 +0000 (09:50 +0100)]
Tidied up a bit

11 years agosort of got the assistant working with the combo model
John Darrington [Sat, 26 Jan 2013 20:01:46 +0000 (21:01 +0100)]
sort of got the assistant working with the combo model

11 years agoBasically working model
John Darrington [Sat, 26 Jan 2013 17:35:05 +0000 (18:35 +0100)]
Basically working model

11 years agoAdded property for the backend
John Darrington [Sat, 26 Jan 2013 11:46:50 +0000 (12:46 +0100)]
Added property for the backend

11 years agoBasic model
John Darrington [Sat, 26 Jan 2013 11:17:27 +0000 (12:17 +0100)]
Basic model

11 years agoBasic Framework for test program
John Darrington [Sat, 26 Jan 2013 09:54:39 +0000 (10:54 +0100)]
Basic Framework for test program

11 years agoAdded a test program for the spread interface
John Darrington [Sat, 26 Jan 2013 09:04:02 +0000 (10:04 +0100)]
Added a test program for the spread interface

11 years agoOnly one spreadsheet
John Darrington [Thu, 24 Jan 2013 19:10:53 +0000 (20:10 +0100)]
Only one spreadsheet

11 years agoformats page: Construction is initialisation
John Darrington [Thu, 24 Jan 2013 18:58:51 +0000 (19:58 +0100)]
formats page: Construction is initialisation

11 years agoFirst line page construction is initialisation
John Darrington [Thu, 24 Jan 2013 18:33:59 +0000 (19:33 +0100)]
First line page construction is initialisation

11 years agoseparators construction is initialisation
John Darrington [Thu, 24 Jan 2013 18:09:47 +0000 (19:09 +0100)]
separators construction is initialisation

11 years agoMove column count from seperators into assist
John Darrington [Thu, 24 Jan 2013 17:58:30 +0000 (18:58 +0100)]
Move column count from seperators into assist

11 years agoMake sheet spec page opaque
John Darrington [Wed, 23 Jan 2013 20:35:03 +0000 (21:35 +0100)]
Make sheet spec page opaque

11 years agoMove all intro related stuff to the intro file
John Darrington [Wed, 23 Jan 2013 18:08:36 +0000 (19:08 +0100)]
Move all intro related stuff to the intro file

11 years agoIntro page: contruction is initialisation
John Darrington [Tue, 22 Jan 2013 19:04:44 +0000 (20:04 +0100)]
Intro page: contruction is initialisation

11 years agoMove file_name member from sri to spreadsheet struct
John Darrington [Mon, 14 Jan 2013 17:42:47 +0000 (18:42 +0100)]
Move file_name member from sri to spreadsheet struct

11 years agoPut the pages as pointers
John Darrington [Sun, 13 Jan 2013 18:43:04 +0000 (19:43 +0100)]
Put the pages as pointers

11 years agoStarted polymorphism of ODS
John Darrington [Sun, 13 Jan 2013 15:01:45 +0000 (16:01 +0100)]
Started polymorphism of ODS

11 years agoStarted polymorphism of Gnumeric
John Darrington [Sun, 13 Jan 2013 12:08:58 +0000 (13:08 +0100)]
Started polymorphism of Gnumeric

11 years agoMade a start at canonicalising the interface
John Darrington [Sun, 13 Jan 2013 09:27:28 +0000 (10:27 +0100)]
Made a start at canonicalising the interface

11 years agoUse convert cell range instead of doing it ourselves
John Darrington [Sat, 12 Jan 2013 14:36:29 +0000 (15:36 +0100)]
Use convert cell range instead of doing it ourselves

11 years agoEnable specification of the cell range
John Darrington [Sat, 12 Jan 2013 12:08:58 +0000 (13:08 +0100)]
Enable specification of the cell range

11 years agoReplace _Import with I_mport in GtkBuilder file.
John Darrington [Sat, 12 Jan 2013 11:15:59 +0000 (12:15 +0100)]
Replace _Import with I_mport in GtkBuilder file.

This means that there is only one _I in the File menu.

11 years agoAdded a bit more to the gui
John Darrington [Tue, 8 Jan 2013 21:13:05 +0000 (22:13 +0100)]
Added a bit more to the gui

11 years agofix compile issues
John Darrington [Sun, 6 Jan 2013 18:27:04 +0000 (19:27 +0100)]
fix compile issues

11 years agoFix syntax generation
John Darrington [Sun, 6 Jan 2013 13:35:45 +0000 (14:35 +0100)]
Fix syntax generation

11 years agoTry to load the correct sheet
John Darrington [Sun, 6 Jan 2013 13:10:22 +0000 (14:10 +0100)]
Try to load the correct sheet

11 years agodata import gui: update user interface
John Darrington [Sun, 6 Jan 2013 09:35:46 +0000 (10:35 +0100)]
data import gui: update user interface

11 years agoChange label from "import text" to "import".
John Darrington [Sun, 6 Jan 2013 08:01:20 +0000 (09:01 +0100)]
Change label from "import text" to "import".

Because it is not only used for text now.

11 years agoSpreadsheet readers: Seperate the mandatory from the optional arguments.
John Darrington [Sun, 6 Jan 2013 07:48:24 +0000 (08:48 +0100)]
Spreadsheet readers: Seperate the mandatory from the optional arguments.

11 years agoBasic working spreadsheet import dialog (sort of)
John Darrington [Sat, 5 Jan 2013 19:35:23 +0000 (20:35 +0100)]
Basic working spreadsheet import dialog (sort of)

11 years agoAdded file chooser filters for spreadsheet file types
John Darrington [Sat, 5 Jan 2013 13:21:56 +0000 (14:21 +0100)]
Added file chooser filters for spreadsheet file types

11 years agoSeperate test-data-import-dialog into different files
John Darrington [Sat, 5 Jan 2013 12:35:19 +0000 (13:35 +0100)]
Seperate test-data-import-dialog into different files

11 years agoFixed a bug reading gnumeric files. 20130217030503/pspp
John Darrington [Sat, 16 Feb 2013 13:44:31 +0000 (14:44 +0100)]
Fixed a bug reading gnumeric files.

Importing  gnumeric spreadsheets would assert-fail if there were empty columns at the start of the sheet.

11 years agoExamine vs. Boxplots: Avoid labels overlapping one another 20130214030508/pspp 20130215030508/pspp 20130216030504/pspp
John Darrington [Wed, 13 Feb 2013 12:03:00 +0000 (13:03 +0100)]
Examine vs. Boxplots: Avoid labels overlapping one another

One factored boxplots remove the name of the factor variables, since
these can be inferred from the chart title.  Also trim off any
leading whitespace from the values.  This reduces the chances of the
labels clashing with one another when many boxplots appear on the
same chart.

Closes bug #38132

11 years agoOutput Viewer Export: Automatically append filename suffix 20130213030508/pspp
John Darrington [Tue, 12 Feb 2013 14:07:11 +0000 (15:07 +0100)]
Output Viewer Export: Automatically append filename suffix

When exporting the output viewer using the file chooser, automatically append
a . and a three letter suffix indicating the format of the export.
Closes bug #38133

11 years agoREGRESSION: Added mention of the dependent variable to table titles. 20130210030503/pspp 20130211030502/pspp 20130212030504/pspp
John Darrington [Sat, 9 Feb 2013 16:19:58 +0000 (17:19 +0100)]
REGRESSION: Added mention of the dependent variable to table titles.

Closes #34732

11 years agoFixed compiler warning placement-parser.c 20130206030505/pspp 20130207030503/pspp 20130208030503/pspp 20130209030504/pspp
John Darrington [Tue, 5 Feb 2013 17:39:46 +0000 (18:39 +0100)]
Fixed compiler warning placement-parser.c

11 years agocasereader: Remove casereader_split() function. 20130203030502/pspp 20130204030505/pspp 20130205030507/pspp
Ben Pfaff [Sat, 2 Feb 2013 16:52:54 +0000 (08:52 -0800)]
casereader: Remove casereader_split() function.

It no longer has any users.

Reported by John Darrington.

11 years agoRANK: Add support for temporary transformations.
Ben Pfaff [Fri, 1 Feb 2013 06:02:08 +0000 (22:02 -0800)]
RANK: Add support for temporary transformations.

Bug #37999.
Reported by Zoltan Fabian.

11 years agoRANK: Adopt a new ranking implementation.
Ben Pfaff [Thu, 31 Jan 2013 07:03:29 +0000 (23:03 -0800)]
RANK: Adopt a new ranking implementation.

Before this commit, the implementation of RANK made multiple passes
through the active file, opening and closing it (with proc_open()
and proc_commit()) as many times as there were input variables.
This worked in simple cases, but it could never work with
TEMPORARY since the second proc_open() will see a different set
of data from the first one.

This commit rewrites RANK to open and read the active file only
once.  It does not make RANK properly work with TEMPORARY, but
it brings it much closer.  It may also be faster in some cases
because, although it makes the same number of passes through
the input data (necessarily), each pass discards all the input
columns except the ones that are really need for that pass.

11 years agoRANK: Create all variables together, in order.
Ben Pfaff [Thu, 31 Jan 2013 06:51:02 +0000 (22:51 -0800)]
RANK: Create all variables together, in order.

An upcoming commit will rewrite the RANK implementation so that the
new variables are not created until after a pass through the data.
(This makes sense because their values cannot actually be determined
until that pass is complete, so there is no point in allocating space
for them in cases.)  To do that, it is necessary to figure out the
variable names (and that they will be valid variable names) in
advance.  This commit switches to that approach in advance.

This approach has another small advantage: the order of the variables
added by RANK to the dictionary does not depend on whether the
variables are named by the user or by generating a name.  (This
is why the rank.at test case changes.)

11 years agoRANK: Simplify rank_sorted_file() with new function sum_weights().
Ben Pfaff [Thu, 31 Jan 2013 05:19:53 +0000 (21:19 -0800)]
RANK: Simplify rank_sorted_file() with new function sum_weights().

This makes the code easier to read and possibly even faster.

11 years agoRANK: Simplify fraction_name() function.
Ben Pfaff [Thu, 24 Jan 2013 06:55:24 +0000 (22:55 -0800)]
RANK: Simplify fraction_name() function.

The caller only needs a constant string so we might as well just return
one directly rather than through a static buffer.

11 years agoRANK: Put #include directives into typical order.
Ben Pfaff [Thu, 24 Jan 2013 06:54:22 +0000 (22:54 -0800)]
RANK: Put #include directives into typical order.

11 years agoRANK: Remove write-only struct member 'ascending'.
Ben Pfaff [Thu, 24 Jan 2013 06:52:10 +0000 (22:52 -0800)]
RANK: Remove write-only struct member 'ascending'.

11 years agoRANK: Lowercase the name of "enum RANK_FUNC".
Ben Pfaff [Mon, 21 Jan 2013 23:45:58 +0000 (15:45 -0800)]
RANK: Lowercase the name of "enum RANK_FUNC".

It is fairly unusual to give a type an all-uppercase name.  The name looks
more natural to me in lowercase.

11 years agotransformations: Relax the rules for transformation finalizing.
Ben Pfaff [Tue, 29 Jan 2013 06:54:18 +0000 (22:54 -0800)]
transformations: Relax the rules for transformation finalizing.

The trns_chain data structure has a barely useful concept called
"finalization".  In practice this is used to make sure that control
structures (e.g. DO IF) that are opened get closed (e.g. END IF).  There
are currently some restrictions on finalizing: namely, transformations
can't be added after a chain is finalized.  Since finalizers are barely
used, we can relax this restriction, which this commit does.  This will be
used in an upcoming commit where the ability to add a transformation to a
finalized change becomes useful for a corner case.

11 years agosubcase: New function subcase_add_vars_always().
Ben Pfaff [Fri, 25 Jan 2013 07:16:44 +0000 (23:16 -0800)]
subcase: New function subcase_add_vars_always().

This function will be used in an upcoming commit.

11 years agocasegrouper: Add comments.
Ben Pfaff [Tue, 22 Jan 2013 03:28:57 +0000 (19:28 -0800)]
casegrouper: Add comments.

11 years agoEnsure that RELIABILITY is always fully constructed. 20130129030504/pspp 20130130030502/pspp 20130131030506/pspp 20130201030505/pspp 20130202030503/pspp
John Darrington [Mon, 28 Jan 2013 18:00:29 +0000 (19:00 +0100)]
Ensure that RELIABILITY is always fully constructed.

Commit e94a39ff572a51907545497c26faccdf4b2c5ada added a 'no crash' test
checking that RELIABILITY's destructor didn't cause any problems when
the procedure was presented with invalid syntax.  Unfortunately the
associated fix was only half done.  The scale_name variable was being
destroyed when it hadn't been initialised.  This change fixes that.

Reported-by: Jeremy Lavergne
11 years agoReliability: Fix crash on invalid syntax 20130126030502/pspp 20130127030503/pspp 20130128030503/pspp
John Darrington [Fri, 25 Jan 2013 11:34:31 +0000 (12:34 +0100)]
Reliability: Fix crash on invalid syntax

11 years agoImprove the printing size on Windows. 20130121030513/pspp 20130122030515/pspp 20130123030508/pspp 20130124030506/pspp 20130125030503/pspp
John Darrington [Sun, 20 Jan 2013 13:16:19 +0000 (14:16 +0100)]
Improve the printing size on Windows.

There were reports that on Windoze the printed output was tiny.  I
think this commit might improve things a little.

11 years agoFix xr to point unit conversion in cairo output driver.
John Darrington [Sun, 20 Jan 2013 11:49:54 +0000 (12:49 +0100)]
Fix xr to point unit conversion in cairo output driver.

The conversion between points (1/72") and xr units was wrong.
This meant that some things were slightly the wrong size.

Reviewed-by: Ben Pfaff
11 years agoOutput window: properly handle the dispose/finalisation
John Darrington [Sun, 20 Jan 2013 13:53:41 +0000 (14:53 +0100)]
Output window: properly handle the dispose/finalisation

11 years agoDocument and implement "precision record" in portable file format. 20130117030512/pspp 20130118030503/pspp 20130119030502/pspp 20130120030506/pspp
Ben Pfaff [Thu, 17 Jan 2013 07:32:26 +0000 (23:32 -0800)]
Document and implement "precision record" in portable file format.

11 years agoRemove configure flag --enable-anachronistic-dependencies 20130116030504/pspp
John Darrington [Tue, 15 Jan 2013 18:04:53 +0000 (19:04 +0100)]
Remove configure flag --enable-anachronistic-dependencies

This flag was a kludge and is not used anymore anyway.

11 years agolexer: Generalize lex_match_phrase() to handle any syntax. 20130113030506/pspp 20130114030505/pspp 20130115030506/pspp
Ben Pfaff [Sat, 12 Jan 2013 20:01:16 +0000 (12:01 -0800)]
lexer: Generalize lex_match_phrase() to handle any syntax.

This makes lex_match_phrase() slightly more useful.  It also eliminates
the ASCII-only requirement.

11 years agoscan: Introduce string_lexer for simple tokenizing of a string.
Ben Pfaff [Sat, 12 Jan 2013 19:10:08 +0000 (11:10 -0800)]
scan: Introduce string_lexer for simple tokenizing of a string.

The following commit will introduce a user outside of the tests.

11 years agosegment: Don't require the input to end in a new-line.
Ben Pfaff [Mon, 7 Jan 2013 06:42:53 +0000 (22:42 -0800)]
segment: Don't require the input to end in a new-line.

11 years agosegment: Separate SEG_N_TYPES from enum segment_type.
Ben Pfaff [Mon, 7 Jan 2013 06:41:17 +0000 (22:41 -0800)]
segment: Separate SEG_N_TYPES from enum segment_type.

With SEG_N_TYPES not actually a member of enum segment_type, GCC doesn't
complain if it's missing from a switch statement on that type.

11 years agoidentifier: Make lex_id_get_length() handle Unicode.
Ben Pfaff [Sun, 6 Jan 2013 23:20:32 +0000 (15:20 -0800)]
identifier: Make lex_id_get_length() handle Unicode.

This function's only caller is documented only to handle ASCII, so this
commit does not fix any bug, but it seems better to generalize our code.

11 years agocairo: Include command name in error messages.
Ben Pfaff [Sat, 12 Jan 2013 17:43:18 +0000 (09:43 -0800)]
cairo: Include command name in error messages.

Commit ddb7b52128d8 (output: Make errors, warnings, and notes into a new
"message_item".) changed command name tracking to a responsibility of
individual output drivers, and converted the output drivers to do it.
However, the conversion of the cairo driver was incomplete.  This commit
fixes that problem.

Reported by John Darrington.

11 years agoGnumeric Reader: Set dictionary to NULL on error
John Darrington [Sat, 12 Jan 2013 13:25:51 +0000 (14:25 +0100)]
Gnumeric Reader: Set dictionary to NULL on error

11 years agoSimplify creation of pango layout in xr driver. 20130112030502/pspp
John Darrington [Fri, 11 Jan 2013 14:26:58 +0000 (15:26 +0100)]
Simplify creation of pango layout in xr driver.

Create the layouts with pango_cairo_create_layout instead of
pango_layout_new.  This seems simpler and avoids a kludge.

Reviewed-By: Ben Pfaff
11 years agodoc: Better describe the meaning of THRU in the RECODE command. 20130110030505/pspp 20130110082819/pspp 20130111030503/pspp 20130111085253/pspp
Ben Pfaff [Thu, 10 Jan 2013 05:52:37 +0000 (21:52 -0800)]
doc: Better describe the meaning of THRU in the RECODE command.

pohaku <pg@hawaii.edu> reported that the description was ambiguous.

11 years agodoc: Improve formatting of RECODE command description.
Ben Pfaff [Thu, 10 Jan 2013 05:51:43 +0000 (21:51 -0800)]
doc: Improve formatting of RECODE command description.

11 years agoFixed a bug reading gnumeric files. 20130107030506/pspp 20130108100915/pspp 20130109030502/pspp
John Darrington [Sun, 6 Jan 2013 18:14:32 +0000 (19:14 +0100)]
Fixed a bug reading gnumeric files.

Some gnumeric files use <gnm:Name> tags for miscellaneous purposes.  Our code
had always assumed that it represented the name of the Sheet.  Hence these files
would read the wrong sheet of a workbook.  This change fixes it.  Closes 38028

11 years agoGET DATA: Produce explicit error on invalid sheet index.
John Darrington [Sun, 6 Jan 2013 16:01:34 +0000 (17:01 +0100)]
GET DATA: Produce explicit error on invalid sheet index.

11 years agoGET DATA: Add error message on incorrect syntax
John Darrington [Sun, 6 Jan 2013 15:52:58 +0000 (16:52 +0100)]
GET DATA: Add error message on incorrect syntax

Previously, when an incorrect syntax reading a spreadsheet file was used, the procedure
would silently fail.  Now it fails with an error message.

11 years agoGnumeric reader: Avoid potential crash reading invalid gnumeric files 20130106030502/pspp
John Darrington [Sat, 5 Jan 2013 15:49:53 +0000 (16:49 +0100)]
Gnumeric reader: Avoid potential crash reading invalid gnumeric files

11 years agoidentifier: Broaden the class of characters allowed in identifiers. 20130104030504/pspp 20130105030504/pspp
Ben Pfaff [Fri, 4 Jan 2013 05:34:25 +0000 (21:34 -0800)]
identifier: Broaden the class of characters allowed in identifiers.

It appears that SPSS allows almost any Unicode character in an identifier,
and particular U+00B4 ACUTE ACCENT.  This commit adds more permitted
characters to the identifier checks.

Reported by Helen Barghan <kenny4president@web.de>.

11 years agoexpressions: Fix dependency on current year in tests. 20130102030504/pspp 20130103030502/pspp
Ben Pfaff [Wed, 2 Jan 2013 06:06:59 +0000 (22:06 -0800)]
expressions: Fix dependency on current year in tests.

The tests for expressions broke on Jan 1, 2013 because the default epoch
depends on the current year.  This commit fixes the tests by setting a
fixed epoch for dates.

Reported by John Darrington.

11 years agovariable: Remove 'aux' member from struct variable.
Ben Pfaff [Wed, 2 Jan 2013 02:34:56 +0000 (18:34 -0800)]
variable: Remove 'aux' member from struct variable.

11 years agoperl-module: Drop use of variable aux data.
Ben Pfaff [Wed, 2 Jan 2013 05:36:22 +0000 (21:36 -0800)]
perl-module: Drop use of variable aux data.

The variable aux data interfaces are not very clean, and furthermore
they have few users.  This commit eliminates the last user.

11 years agoperl-module: Put struct dictionary inside a wrapper "struct pspp_dict".
Ben Pfaff [Wed, 2 Jan 2013 02:47:53 +0000 (18:47 -0800)]
perl-module: Put struct dictionary inside a wrapper "struct pspp_dict".

In an upcoming commit this will allow an extra member to be associated
with the Perl version of each dictionary.

11 years agoperl-module: Rename sysfile_info to syswriter_info.
Ben Pfaff [Wed, 2 Jan 2013 03:18:31 +0000 (19:18 -0800)]
perl-module: Rename sysfile_info to syswriter_info.

This module had sysfile_info and sysreader_info.  The former was a writer,
the latter a reader.  I found the asymmetric names a little confusing, so
this commit renames them more consistently.

11 years agocase-map: Drop use of variable aux data.
Ben Pfaff [Wed, 2 Jan 2013 02:30:39 +0000 (18:30 -0800)]
case-map: Drop use of variable aux data.

The variable aux data interfaces are not very clean, and furthermore
they have few users.  This commit eliminates one of the users.

11 years agoCROSSTABS: Drop use of variable aux data.
Ben Pfaff [Wed, 2 Jan 2013 03:15:16 +0000 (19:15 -0800)]
CROSSTABS: Drop use of variable aux data.

The variable aux data interfaces are not very clean, and furthermore
they have few users.  This commit eliminates one of the users.

11 years agohmap: New interfaces for iterating a bucket without comparing hashes.
Ben Pfaff [Fri, 27 Aug 2010 05:49:38 +0000 (22:49 -0700)]
hmap: New interfaces for iterating a bucket without comparing hashes.

11 years agoUpdate tests to reflect change in EXAMINE / EXTREME behaviour.
John Darrington [Tue, 1 Jan 2013 17:32:03 +0000 (18:32 +0100)]
Update tests to reflect change in EXAMINE / EXTREME behaviour.

Commit bd156adaff5b7c1bbe48b5c64006ead58d9a37d6 slightly changed the
behaviour of the EXTREME subcommand of the EXAMINE procedure, but the
tests did not reflect this.

This change updates the tests accordingly.  Thanks to Zoltan Fabian
for confirming which was the correct behaviour.

11 years agoFix bug #37984 - EXAMINE extremes vs. fractional weights.
John Darrington [Tue, 1 Jan 2013 15:19:26 +0000 (16:19 +0100)]
Fix bug #37984 - EXAMINE extremes vs. fractional weights.

There was a bug where extreme values were not calculated properly when
weights were fractional.  This change fixes this problem and adds a
test.

11 years agoRemove assertions which compare the sum of weights between passes. 20130101030509/pspp
John Darrington [Mon, 31 Dec 2012 10:33:38 +0000 (11:33 +0100)]
Remove assertions which compare the sum of weights between passes.

These asserted that the sum of case-weights of a dataset calculated
in one pass, was the same as that calculated in a second pass.
Algebraically this is correct.  However, for optimisation purposes,
it is sometimes desireable that the second pass occurs after the
data has been reordered.   If that happens, the sum of weights can
be slightly different due to floating point rounding errors.  This
happens particularly when the caseweights are fractional.

11 years agocsv-file-writer: Fix implementation of decimal point option. 20121228030502/pspp 20121229030503/pspp 20121230030503/pspp 20121231030505/pspp
Ben Pfaff [Fri, 28 Dec 2012 05:17:43 +0000 (21:17 -0800)]
csv-file-writer: Fix implementation of decimal point option.

Now that pspp sets LC_NUMERIC, dtoastr() might yield either '.' or ',' as
the decimal point, so the CSV writer needs to check for either one and
replace it by the decimal point requested by the caller.

Reported by John Darrington.

11 years agoi18n: New functions for UTF-8 case conversion.
Ben Pfaff [Fri, 28 Dec 2012 03:51:08 +0000 (19:51 -0800)]
i18n: New functions for UTF-8 case conversion.

Also, use the new functions in a few cases where we want a full UTF-8
conversion.

11 years agoUse UTF-8 case-insensitive hashes and comparisons for language identifiers.
Ben Pfaff [Fri, 28 Dec 2012 03:32:40 +0000 (19:32 -0800)]
Use UTF-8 case-insensitive hashes and comparisons for language identifiers.

The PSPP language has case-insensitive identifiers (variable names, etc.)
but until now it has only implemented case insensitivity for ASCII
characters.  This commit properly implements case insensitivity for all
Unicode characters, using libunistring.

Bug #31072.