GnuCash 2.3.11 released
Phil Longstaff
plongstaff at rogers.com
Fri Mar 19 07:51:24 EDT 2010
Announcement: GnuCash 2.3.13 (Unstable) Release 2010-03-19
GnuCash 2.3.11 (Unstable) released
The GnuCash development team proudly announces GnuCash 2.3.11, the
twelfth of several unstable 2.3.x releases of the GnuCash Free
Accounting Software which will eventually lead to the stable version
2.4.0. With this new release series, GnuCash can use an SQL database
using SQLite3, MySQL or PostgreSQL. It runs on GNU/Linux, *BSD, Solaris,
Microsoft Windows and Mac OSX.
WARNING: This is an *UNSTABLE* version of Gnucash.
This release is intended for developers and testers who want to help
tracking down all those bugs that are still in there.
Make sure you make backups of any files used in testing versions of
GnuCash in the 2.3.x series. Although the developers go to great lengths
to ensure that no data will be lost we cannot guarantee that your data
will not be affected if for some reason GnuCash crashes in testing these
releases.
NOTE: The latest stable version is 2.2.9.
PLEASE TEST TEST AND TEST SOME MORE any and all features important to
you. Then post any bugs you find to bugzilla
Major changes in the 2.3.x release include;
* In addition to the XML backend, Gnucash can now use a SQLite3,
MySQL or PostgreSQL database to store the data. This is a new
implementation using libdbi. It supports all features including
the business features. In order to build with this, add
--enable-dbi to the configure command. In addition to the
libdbi-dev package for your distribution, you will also need the
appropriate DBD (libdbi driver) package for sqlite3, mysql or
postgresql.
* As a replacement for the current GtkHTML HTML engine used to
display reports and graphs, Gnucash can use WebKit. WebKit is
the engine used by Google Chrome on Windows and Safari on Apple.
In order to build with this, add --enable-webkit to the
configure command. You will need an appropriate webkit-dev
package. On win32, you will need to download the
webkit-1.1.5-win32.zip file from the source repository and put
it into the downloads directory of your gnucash build area.
* Updated to AqBanking 3 on Win32.
Changes between 2.3.10 and 2.3.11 include:
* Bugs fixed
* Partial fix for #611014 - CSV import crashes Bug #611014
mentions a problem where after setting the columns,
clicking OK and selecting a target account, the Date
column is lost. This is because of a difference in
behaviour on windows and linux. The code attaches to the
"edited" signal of the renderer. On linux, this signal
is emitted when a new combo box item is selected. On
windows, the combo box needs to be selected and focus
lost before the signal is emitted. This is changed to
the "changed" signal of the renderer which acts as
expected on both platforms.
* Bug #140400 - Crash when deleting an account that is
still in use by the business features Add a dialog which
contains a list of objects referring to the account and
an explanation that these need to be deleted or have the
account reference modified. Note: I'm no UI designer.
This is functional, but if anybody wants to pretty it
up, go ahead.
* Bug #536108 - After cancelling "save as", user is not
prompted 2nd time
* Bug #507351 Terms not defined prior to use This commit
changes the text in the accounts druid to explain what a
placeholder account is, together with some additional
improvements in the wording of that page. Since this
increases the amount of text, the width of the label has
been modified as well.
* Fix Bug 590570 - When deleting more than one report in
sequence the program crashes Delete the custom report
backup file before renaming the current custom report
file.
* Fix Bug 611346 - Crash in Saved Reports dialog if you
select Delete button with no report selected Test for no
report selected before trying to use the selected report
guid.
* Bug #364358 - Import dialog unreadable with dark colored
gtk theme (with solution)
* Bug #525821 new or edited account names should be
checked for reserved chars like ":"
* Bug #610675 Register Tabs Do Not Display Since Nightly
Build r18685 Apply Bob's patch after fixing the
whitespace. This patch may cause issues on Windows (a
previous version of the patch did). If the next Windows
nightly build exhibits the missing register tab names
again, it will have to be reverted again and improved
upon.
* Bug #611645 gnc-svnversion returns "too long" a string
for git users, patch by Jeff Kletsky
* Bug#325436 creating income account for invoice doesn't
restrict account type
* Fix Bug 611810 - GC crashes when I click on File ->
Properties SCM files still refered to gnc-book-get-slots
which had been removed. gnc-book-get-slots was replaced
everywhere by qof-book-get-slots.
* Fix bug 611885 - Crash when opening postgresql file
Previous work setting Timespec values via gobject
properties missed the case where the timespec loaded
from the database was NULL.
* Bug #611140: Fix crash on Open Subaccounts (hopefully).
* Bug #610321 - Compile errors with gtk-2.10.4:
GTK_BUTTONBOX_CENTER undeclared This commit applies a
reworked version of the patch to remove the use of
GHashTableIter by Cristian Klein. Cristian's patch
didn't apply cleanly to the current trunk. I have made
the necessary changes to make it apply (and hopefully do
what Cristian intended).
* Bug #611470 Add Japanese concepts guide into GnuCash
installer, patch by Yasuaki Taniguchi
* Bug 605802: Can't input Japanese characters at an
account register window on windows with SCIM, IIIMF and
XIM Latest patch by Yasuaki Taniguchi to fix two
problems 1) Can't use account separator char when
entering account name in a split, and 2) Can't use
<SHIFT>+ and <SHIFT>- to go forward/backward a week.
* Fix bug 591177: Printer font is too small to read with
webkit as html renderer. From comment 23: "The PDF in
comment 2 is about 8 times smaller than it should be.
One possible cause of this bug is if
gtk_print_operation_set_unit (op, GTK_UNIT_POINTS) is
not called. gtkprint defaults to GTK_UNIT_PIXEL - a
useless unit to be using with printers. On Linux
GTK_UNIT_PIXEL units are 1 unit = 1/72 inch (the same as
GTK_UNIT_POINTS as well as PostScript and PDF units). On
Windows GTK_UNIT_PIXEL units are the GDI device units
which for printers is the dpi resolution. So for a
600dpi printer 1 unit is 1/600". If the application was
developed on Linux and assumes the default gtkprint
units are always 1/72" inch the output on Windows using
a 600dpi printer will be 72/600 = 0.12 of the size (or
approximately 1/8 of the size)." Solution was to use
webkit_web_frame_print_full() which allows us to provide
our own GtkPrintOperation object with units set to
GTK_UNIT_POINTS. Tested on both Linux and Windows.
* Bug #610675: Revert the gnc-main-window parts of r18637
because it makes the tab names disappear under Windows.
This disables the enhancement of bug#608329 again, but
the disappearing of the tab names is a rather major bug.
Note that we either need to fix the enhancement again,
or revert the rest of r18637 as well so that the account
properties don't show the color chooser without any
effects.
* Bug #605802: Fix input of Japanese characters in
register with SCIM, IIIMF and XIM Patch by Yasuaki
Taniguchi. Revised and extended version of r18638. The
main functions are as follows: (1) synchronization of
preedit string between the register window and
sheet->entry, (2) application to pango attributes to
preedit string in the register window, (3) include
scroll offset patch (id=153514), (4) include preedit
string rollback patch (id=153518), (5) fix formula and
account cells input problem which Christian pointed out,
(6) surpress quick-fill when preedit string exists, (7)
fix Windows IME problem. (8) Fix quick-fill problem.
* Redo of the dot-gnucash fix (so that GNC_DOT_DIR
actually works) fixing bug 610707, adding Doxygen
comments for all functions in gnc-filepath-utils, and
adding testing for xaccResolveURI and more tests for
xaccResolveFilePath.
* Revert r18713 (reopen 605802 "Input of Japanese
characters". This commit had 2 problems: 1) when
entering an account name, the account separator would no
longer accept at the current level of the account tree
and move to the next level 2) <SHIFT>+ and <SHIFT>- in a
date field would not change the field by 1 week.
* Bug #610348: Add compiling our own binary of libbonoboui
because the ftp.gnome.org binary still depends on the
obsolete libxml2.dll. However, we still download the
erroneous binary and unpack it into $GNOME_DIR because
libgnomeui depends on libbonoboui which in turn depends
on libgnome. Hence, libbonoboui cannot be compiled
before libgnome-dev is unpacked, but libgnomeui won't
report to be installed correctly before libbonoboui is
available as well. Theoretically, we would have to split
the inst_gnome step so that it first unpacks libgnome et
al., then we run the inst_libbonoboui step, then we run
the second part of inst_gnome which would be something
like inst_gnomeui. I'm lazy, so I silently overwrite the
libbonoboui DLL with our hand-compiled version and
that's it.
* Bug #608032 - MySQL timeout and no attempt reconnect,
second version This version builds on Phil's
implementation of the dbi error callback functions to
test for a timeout and to do the reconnect. The same
error handling is equally implemented for postgres and
sqlite. Unlike MySQL these two database types don't
actually generate timeouts, but the functionality can be
used later on for other error types as well.
* Bug #610051 - Crash when using GktHtml whenever a report
is opened
* Bug #610348 missing dependencies in windows build Update
gnome-vfs to 2.24.1 Note that this means Windows 2000 is
no longer supported.
* Partial fix for bug #610321 Compile errors with
gtk-2.10.4: GTK_BUTTONBOX_CENTER undeclared
GTK_BUTTONBOX_CENTER is replaced with
GTK_BUTTONBOX_START
* Fix bug #564380 additionally for easy invoice and fancy
invoice. Patch by Mike Evans.
* Bug #610047 - Dutch accounts template doesn't work Add
missing closing brackets
* Bug #609044: Improve UI strings for tax report options
Patch by Frank H. Ellenberger: As we currently have a
nice US income and a partial german VAT tax report, I
feel uncomfortable with the change of r18413, which
renamed Edit->Tax options to Income Tax options. So here
is another approach, to clarify the tax report and
business tax menu points.
* Bug #609043: Improve (mostly german) translation of txf
Patch by Frank H. Ellenberger: This patch is a first
extract of so an approach, which I have here, which will
probably lead to a german income tax declaration ESt-A1.
But this extract is more general and based on r18413
changes. Changes: 1. Header Comments: adding
Contributors 2. Most strings in de_DE translated to
german.
* Bug #608032: Handle MySQL connection timeouts with
reconnect Patch by Tom van Braeckel: For the full
discussion, see the mailing list:
https://lists.gnucash.org/pipermail/gnucash-devel/2010-February/027323.html Rationale: When we try to open a database transaction, and the database reports that the "server has gone away", we try to reconnect before failing hard.
* Bug #609005: Add recipient name on invoices Patch by
Mike E: Having set up a client/customer including the
name of a recipient when I print an invoice the
recipients name ("Account Dept" say) is not printed in
the invoice. I think this is a bug rather than a
feature. I have attached a patch to fix. It still prints
the company name above the recipient name however. I
could submit an additional patch to provide an invoice
option to toggle printing of the company name if
users/developers feel they want this option, as I do.
* Bug #609603: Windows packing/win32/install.sh PATH fix
Patch by Yasuaki Taniguchi: When I
run /c/soft/gnucash/inst/bin/gnucash or gnucash.cmd
after I finish building win32 binary, DLL missing dialog
boxes pop up. Missing DLLs are - libgcrypto.dll, -
libPQ.dll, - mysql.dll, and - ssleay32.dll. This patch
add search path to them to fix this problem.
* Bug #564380 Payment on bills doubles bill Patch by Mike
Evans
* Updated translations or translation-related changes
* Updated Japanese translation, copied from the
Translation Project.
* Updated Persian (Farsi) translation by Mehdi Alidoost.
* Add dutch translation to the Windows installer.
* Updated Slovak translation, copied from the Translation
Project.
* Updated Dutch translation, copied from the Translation
Project.
* Update German translation.
* Add implementation of Gregorian - Jalali converter code.
Copied from http://www.farsiweb.ir/wiki/Iranian_calendar
* Other user-visible changes
* Change file loading message to "Loading user data..."
Reading file is technically only correct for files not
for databases.
* Display the SX variables in the "Since Last Run..."
dialog in ASCII order rather than ordering by hashvalue
(i.e. no order at all) as was done previously. Patch by
Jesse Weinstein.
* Add bzr support to gnc-svnversion. Patch by Jesse
Weinstein
* Fix crash on opening the tax report, introduced by
r18673. Patch by Alex Aycinena.
* Revert r18881/18884/18885 (bug #610675 - Register Tabs
Do Not Display Since Nightly Build r18685)
* Fix missing color in qif-import account copy, patch by
Tom Van Braeckel
* Replace Income Tax Options with Tax Report Options.
Patch by J. Alex Aycinena.
* Patch my patch in r18884. Spotted by Herbert Thoma.
* Regression fix: patch for colored tabs segfault in
trunk, patch by Tom Van Braeckel
* Other code/build changes
* Remove test-load-module from tests
* Fix Makefile.am handling of gncla-dir.h so that it will
be removed so that 'make distcheck' will pass
* Fix guint32 vs gint32 in gnc-uri-utils test.
* Update POTFILES.in based on new source files
* Complete unit tests for gnc-uri-utils api and fix bug
found by running the tests.
* More doxygen.log-prompted typo fixes, patch by Jesse
Weinstein
* Fix typo in previous commit
* Add first test for the gnc-uri-utils api. This test
verifies gnc_uri_get_components.
* Implement the object reference infrastructure routines
to allow a list of business objects referring to a
specific other object (e.g. an account) to be
determined. This will help fix bug 140400 because the
account delete code can now determine a list of business
(or other) objects which have references to that
account, and prevent the account from being deleted
while references still exist.
* Add some new gobject-related infrastructure so that when
deleting an object, it can be determined if there are
other objects with references to that object (bug
140400). Some routines are normal routines, and some
routines use the gobject structure to allow different
implementations by different object types. Per-instance
routine: gboolean
qof_instance_refers_to_object(QofInstance* A,
QofInstance* B) - returns TRUE/FALSE whether object A
contains a reference to object B. Normal routine: GList*
qof_instance_get_referring_object_list_from_collection(QofCollection* c, QofInstance* B) - Calls qof_instance_refers_to_object() for each object in the collection, and returns a list of the objects which contain a reference to B. Per-instance routine: GList* qof_instance_get_typed_referring_object_list(QofInstance* A, QofInstance* B) - returns a list of all objects of the same type as A which contain a reference to B. Being per-instance allows an object to use knowledge to avoid scanning cases where there are no references (e.g. billterms do not contain references to splits), or a scan is not needed (references from splits to an account can be determined using xaccAccountGetSplitList()). This routine can do a scan by calling qof_instnace_get_referring_object_list_from_collection(). Normal routine: qof_instance_get_referring_object_list(QofInstance* A) - For all collections in the book, gets an instance and calls its qof_instance_get_typed_referring_object_list() routine, concatenating all of the returned lists. This is the routine that external code can call to get a list of all objects with references to an object A. The list must be freed by the caller but the contents must not. Per-instance routine: gchar* qof_instance_get_display_name(QofInstance* A) - returns a string which can identify object A to the user. This could be used to display a list of the objects returned by qof_instance_get_referring_object_list() ("Invoice 0004 for customer C") so that the user can modify those objects to remove the references. Note that this is going to require internationalization, which has not been implemented yet. If not overridden by the object class, the default string is "Object <type> <address>" e.g. "Object gncCustomer 0x12345678".
* Add event registration and handling from the
qof_event_handlers. This way, the split view is updated
correctly even on undo/redo.
* Revert r18869, "Move gnc-ui-util.[hc] and gnc-euro.[hc]
from app-utils to engine" Those files indeed belong
better into app-utils; the app-utils defines several
not-yet-GUI parts of gnucash, like many conversions from
and to strings (more than those in these two files), so
these files are well suited in here.
* MSVC compatiblity: Fix include files in MSVC. Replace
QSTRCMP by g_strcmp0 on MSVC.
* Improve non-gnome UI compatiblity: app-utils can compile
without gtk as well.
* Add src/calculation and src/app-utils to cmake build
system. app-utils is needed for the conversion from and
to string for gnc_numeric, date, and other values.
* Fix failing core-utils tests I simply removed the
obsolete test cases. More work is still needed to add
new valid tests.
* Remove some unused variables.
* Win32/MSVC compatiblity - Replace trunc() by floor() -
Provide a round() workaround implementation for MSVC -
Use g_strcasecmp instead of the libc one - Add include
for libc replacements
* Some more const-correctness in engine functions.
* Update POTFILES.in for the moved dialog-userpass.c and
the newely added gnc-jalali.c
* Move gnc-ui-util.[hc] and gnc-euro.[hc] from app-utils
to engine because they don't depend on gtk but are
important additions for the engine types. In particular,
the formatting of a gnc_numeric is defined there.
* Fix circular dependency between gnome and gnome-utils
introduced in r18842
* MSVC compatiblity: snprintf is required to have a
prefixing underscore. Also, more symbols of
libguile/gc.h need explicit declspec on MSVC.
* Add variant of gnc_engine_init which is suitable for
statically the linked-in library.
* Only use long long format specifiers if available -
avoids error message when compiler thinks they're not
available.
* Fix parent/child relationships in billterms in case the
parent hasn't been loaded yet. Remove child column from
billterm table because it duplicates info in the parent
column and just complicates loading objects.
* Fix parent/child links for tax tables. If a tax table's
parent table has not been loaded yet, remember the
relationship, and after all tables have been loaded, fix
up the rest of the parent/child links.
* Handle NULL string pointer as a NULL guid
* Add a mechanism so that the business sql backend module
can provide the main sql backend with the order in which
objects should be loaded. This will allow billterms and
taxtables to be loaded before objects which contain
references to those objects.
* MSVC compatiblity: Replace QOF_BOOK_LOOKUP_ENTITY macro
by a RETURN_ENTITY macro and add inline functions for
lookup. MSVC doesn't accept the syntax with an inlined
block, x = ({ foo; bar; value;}). Hence, this is being
replaced by actual function definitions, and the body of
those functions is defined by the new macro.
* Delete unused variables.
* Use a normalized uri format internally to refer to data
stores. Data stores for GC can be a file (xml or
sqlite3) or a database one some server (mysql or
postgres). Wherever it makes sense internally, data
stores will be referred to via a normalized uri:
protocol://user:password@host:port/path Depending on the
context and story type some of these parts are optional
or unused. To achieve this, a new utility interface has
been setup: gnc_uri_<xxx>_<yyy> that can be used to
manipulate the uris or convert from non-normalized
formats to normalized and back. For example, when the
user selects a file in the Open or Save As dialog,
gnc_uri_get_normalized_uri will convert the file into a
normalized uri. Or when the actual filename is needed
this can be extracted with gnc_uri_get_path. You can
also test if a uri defines a file or something else with
gnc_uri_is_file_uri. For the complete documentation, see
src/core-utils/gnc-uri-uitls.h This commit installs
gnc-uri-utils and modifies the source where it makes
sense to use its convenience functions. This concerns
all functions that had to deal with file access in some
way or another, the history module and the functions
that generate the history menu list and the window
titles. Note that gnc-uri-utils replaces
xaccResolveFilePath and xaccResolveUrl in all cases.
xaccResolveUrl has been removed, because gnc-uri-utils
fully replaces its functionality. xaccResolveFilePath is
used internally in gnc-uri-utils to ensure an absolute
path is always returned (in case of a file uri, not for
db uris). But it has been renamed to
gnc_resolve_file_path to be more consistent with the
other functions. Lastly, this commit also adds a first
implementation to work with a keyring to store and
retrieve passwords, althoug
* Make business backend initialization functions
accessible when statically linking
* MSVC compatibility: Disable "C99 designated
initializers" by a compiler-dependent macro Same as
r18755.
* Make the backend initialization functions accessible
when statically linking
* Change the definition of QOF_STDOUT The old definition
file: conflicts with normal uris that can start with
file: as well. I have chosen > instead, which is never a
valid filename and on unixlike systems associated with
standard out operations.
* Use proper qof CFLAGS/LDFLAGS since core-utils now uses
qof
* The webkit used on win32 has
webkit_web_frame_print_full() defined in include files,
so we don't need a potentially conflicting extern
declaration.
* If gmtime_r is defined as a macro, undef it
* Add svn:ignore to src/core-utils/test
* Remove invalid target (got copied from src/engine/test)
* Build test-core before core-utils
* Add the core-utils tests in the automake system
* Move filepath related tests to core-utils/test
* Undefine localtime_r as a macro (new mingw pthreads
package defines it)
* Add braces to make if-if-then-else structure clear and
avoid GCC 4.4.0 error message
* Remove gncmod-test from test-core It's not used and
prevents test-core from being included in core-utils
tests
* Add missing NULL sentinel at end of g_strconcat()
function call
* Removed erroneously re-created
src/engine/gnc-filepath-utils.c
* Move binreloc library include
* Note the moved files in POTFILES.in.
* Remove C executable from cmake as it is no longer
necessary.
* Add missing link library after r18811.
* Adapt cmake to the file move in r18811.
* Add src/engine/test/test-resolve-url
* Win32: Add download of cmake, disabled by default.
* MSVC compatibility: strftime() doesn't know "%T" here.
Also, g_fopen doesn't work, but fopen does.
* Move gnc-filepath-utils and dependencies from engine to
core-utils
* Typo fixes, as found by doxygen.log, patch by Jesse
Weinstein
* Tweak to gnc-svnversion's bzr section, patch by Jesse
Weinstein
* Use "template-account" property to get/set template
account.
* Add "template-account" to schedxaction as a gobject
property.
* Simplify handling of sx template_acct column.
* Rename GNCBook into QofBook everywhere and remove
separate header gnc-book.h. The former was already
#define'd on the latter, so its removal gets rid of one
level of indirection which makes function lookup easier.
Also, the macro (!) qof_book_get_slots was turned into a
normal function again because that's what functions are
for (and otherwise the additional declaration in
engine.i would break).
* Decrease compiler warnings by removing unused variables.
* Reduce compiler warnings by replacing strerror() with
g_strerror() and similar glib replacements.
* Win32: Add more header includes where necessary to avoid
using undeclared functions.
* Comment out unimplemented function. Improve
const-correctness.
* Include gtk/gtk.h instead of gtk/gtkclist.h as
recommended by Gtk
* Fix make dist r18765 introduced test target
test-resolve-url, but no source file test-resolve-url.c
Removed the target.
* Re-indentation of source code, next batch:
src/gnome-utils/* This also strips trailing whitespaces
from lines where they existed. This re-indentation was
done using astyle-1.24 using the following options:
astyle --indent=spaces=4 --brackets=break --pad-oper
--pad-header --suffix=none
* Re-indentation of source code, next batch:
src/register/* This also strips trailing whitespaces
from lines where they existed. This re-indentation was
done using astyle-1.24 using the following options:
astyle --indent=spaces=4 --brackets=break --pad-oper
--pad-header --suffix=none
* Re-indentation of source code, next batch:
src/business/* This also strips trailing whitespaces
from lines where they existed. This re-indentation was
done using astyle-1.24 using the following options:
astyle --indent=spaces=4 --brackets=break --pad-oper
--pad-header --suffix=none
* Re-indentation of source code, next batch: src/engine/*
This also strips trailing whitespaces from lines where
they existed. This re-indentation was done using
astyle-1.24 using the following options: astyle
--indent=spaces=4 --brackets=break --pad-oper
--pad-header --suffix=none
* Re-indentation of source code, next batch. This also
strips trailing whitespaces from lines where they
existed. This re-indentation was done using astyle-1.24
using the following options: astyle --indent=spaces=4
--brackets=break --pad-oper --pad-header --suffix=none
* Replace and-let* in scheme script so that srfi-2 isn't
needed.
* Replace one more g_list_append by g_list_prepend.
* MSVC compatibility: Somehow fdopen() doesn't work during
the trace file initialization. Use fopen() instead.
* Win32 build: libguile needs minor tweaking for MSVC
compatibility.
* MSVC compatibility: Fix r18748, r18761 by replacing
strncasecmp() with strnicmp().
* MSVC C++ compatibility: Rename the internal name of
union _GUID because _GUID is a builtin keyword here.
This does not concern the typedef name, only the
internal union name, so it doesn't harm us.
* C++ compatibility: namespace is a keyword, so don't use
it as variable name.
* C++ compatibility: export is a keyword, so don't use it
as a member name.
* Add support for passing a Timespec as a boxed GValue
* Fix test makefiles. Many tests now need to include other
libraries because files have changed directories.
* Add more gobject property definitions to GNCPrice,
Transaction, SchedXaction and Split.
* Win32: Create the MSVC import library for libguile
during install.sh.
* Remove static current_session variable of libqof - we
keep one in gnc-session.c already.
* Doxygen improvements, patch by Jesse Weinstein
* More changes where SQL backend uses gobject properties
to load/save objects.
* More conversion to read/write objects from sql backend
using gobject parameters
* Start to add properties to business objects. Currently
only 1 property per object, but this infrastructure will
allow more generic importing of objects.
* Take advantage of the initial property definition for
Transaction, Split, GNCPrice and SchedXaction by
replacing custom access routines with gobject
properties.
* Add a few gobject properties to some engine object
types. This adds more of the gobject infrastructure to
Transaction, Split, SchedXaction and GNCPrice. Gobject
properties provides a standardized interface to the
engine objects which should allow standard and
simplified read/write mechanisms. For the sql backend,
for example, db columns can be mapped to properties. In
a generalized csv importer, csv columns can be mapped to
properties.
* Partly revert r18748, "Win32 compatibility": lib/libc
directory doesn't have glib available.
* Add a "make indent" target, but watch out with using its
result. The "astyle" indent tool unfortunately behaves
significantly different in its different versions (1.22,
1.23, 1.24) even with identical options. That is, the
basic indentation is the same, but in a large project
such as ours, there are just too many special cases
which astyle doesn't get identical due to its bugfixes
and new features. Hence, please use the result of this
target with great care, and if in doubt, just use it for
your own amusement but don't commit the resulting
changes. Thanks!
* Fix up some doxygen comments
* Don't include gnc-lot-p.h where not needed
* Win32: Fix libbonoboui compiling.
* MSVC compatibility: Microsoft doesn't have C99
"designated initializers". Those were introduced in
r17724, bug#539957, but apparently this C99 is not
supported by MSVC and won't be for some time to come.
Hence, for MSVC we need the workaround to define a macro
that will shadow the member names. However, the
initialization itself works fine and non-MSVC code is
unchanged, so I think we can live with that.
* MSVC compatiblity: Struct initialization doesn't work as
expected. Somehow, the struct initialization containing
a gnc_numeric doesn't work. As an exception, we
hand-initialize that member afterwards.
* MSVC compatiblity: open() flags and S_ISDIR doesn't
exist on MSVC.
* MSVC compatiblity: Use a char* pointer for the memcpy()
input argument. This is required by MSVC because we do
some pointer arithmetic in the memcpy() argument, but in
order to do this, MSVC wants to know the pointed-to type
of the pointer because pointer arithmetic increases the
pointer not by a number a bytes but a number of
sizeof(type). MSVC thinks for void* it doesn't count
bytes. We achieve the desired effect by using a char*
pointer so that bytes are counted.
* MSVC compatiblity: Add defines for functions/types which
are available under different names in MSVC.
* MSVC compatiblity: Array initialization in MSVC requires
a constant, not a variable. That is, gcc accepts a
constant variable in many cases now, but MSVC doesn't
accept it. So it must be turned into an old preprocessor
define.
* MSVC compatibility: Remove forward declaration of static
array by reordering the function that uses it.
* Win32 compatibility: Use glib wrappers of non-usual
POSIX functions.
* More header include compatibility: Watch out for
HAVE_UNISTD_H.
* Re-indentation of source code, next batch. This also
strips trailing whitespaces from lines where they
existed. This re-indentation was done using astyle-1.24
using the following options: astyle --indent=spaces=4
--brackets=break --pad-oper --pad-header --suffix=none
* Convert GNCLot to use more gobject features. Removes all
direct access to lot object fields, which are now
accessed through functions or property names (for
backend sql load).
* Modify POTFILES.in to handle source files moved to a new
directory
* Also build backend/xml in cmake.
* Reverted 18699
* Clarify required steps to setup windows build
environment. The mingw website has changed quite a lot
since the README was written and some of the url's used
in it were confusing. I have changed the url's to point
to the actual packages on sourceforge (current at the
time of this writing) and added some extra details where
I had trouble understanding the actions to perform.
* Doxygen fixes - Have this file show up under module
"Utility Functions" - Normalize the function
descriptions (some were not in doxygen format) - Add a
global file description - rename parameter 'file' to
'filename' for better consistency (note this required an
internal parameter to be renamed from filename to
new_filename)
* More minor MSVC code fixes. However, this code doesn't
compile with MSVC9.0 for a few reasons: 1. libguile.h
comes with its own scmconfig.h which contains
HAVE_STDINT_H whereas MSVC doesn't have that. This is
stupid guile which doesn't accept the fact that the user
uses a different compiler than how they compiled guile.
2. Some initializations are not supported:
Account.c:3312 etc. 3. The C99 "designated initializers"
of e.g. Account.c:4661 ff. (r17724, bug#539957) are not
supported by MSVC9.0 - this is the hardest problem of
these all.
* Make CMake system more complete so that it builds on
win32/mingw. Also, add a test executable to check that
we got all the library dependencies.
* More CMake work: Build swig wrappers correctly. Build
gnc-module.
* Cosmetic: Remove duplicate include of gnc-engine.h
* Remove trailing whitespace
* Minor doxygen change and lots or trailing whitespace
removed
* Remove GNOME_PRINT_{CFLAGS,LIBS} from the Makefile
templates
* Spelling errors and trailing whitespace removal
* Remove reference to gnome print in the comments GnuCash
no longer uses gnome print. It has been replaced with
gtk print.
* Remove two more popt references in support files.
* Remove popt requirement from configure GnuCash doesn't
use it. At the same time, I removed a check that has
been commented out since the beginning of the revision
history (somewhere in 2007). This check tested for the
presence of popt.h to then run some libtool changes. As
I said this whole block has been commented out since the
beginning of time, so I considered it to be obsolete,
more even so now that the popt requirement has been
removed.
* Remove popt references - popt has been replaced with the
GOption infrastructure. So there's no need to include
the popt.h file. - Also rewritten the comment that was
elaborating on popt vs GOption - Finally removed the
loglevel option (which was excluded from the compile
anyway) that still referred to popt.
* Fix minor spelling errors
* When creating lists of database objects, use
g_list_prepend() rather than g_list_append(). There may
be cases where the list order is significant and thus
needs to be reversed, but that is not true in these
cases. This provides a large improvement in database
loading performance. Analysis and basis patch supplied
by Donald Allen.
* Update documentation references to Active Perl
(5.8->5.10)
* Remove redundant entries in EXTRA_DIST
* More experimental cmake building. Except for the scheme
wrappers this seems to work until at least the engine
module. However, I didn't tackle the issue with the
generated headers which contain some installation paths
- but those we should get rid of anyway.
* Move two gtk-dependent files from src/core-utils to
src/gnome-utils. core-utils depends on glib and
additionally guile and gconf, but not (yet) gtk. Those
two files which do are moved to the next module which
already depends on gtk, which is gnome-utils.
* Add some experimental CMakeLists.txt which can compile
the libqof part, on Linux, Windows/mingw and (no joke)
Windows/MSVC. I'm interested in some tests with the
cmake build system, but if it doesn't prove useful I
will remove it again within a few weeks.
* Make libqof compatible for MSVC9.0 compiler (no joke).
The main change is that the syntax for variadic macros
is slightly different in MSVC compared to gcc. But they
exist, so offering the log macros in the different
syntax is sufficient.
* Make more header inclusions conditional on whether they
exist.
* Update .gitignore, proposed by Jeff Kletsky
* Make sure file urls actually contain path information or
are NULL
* Remove redundant GLIB check.
* Bump minimum required versions of gtk+, goffice and
gtkhtml gtk+: 2.10 goffice: 0.5.1 gtkhtml: 3.14
* Add a starter script for gnucash under ddd (a gui
frontend for gdb)
* Build fixes for EL5 after glib 2.12 requirement
* Ensure that GNC_DOT_PATH and other gnc_dotgnucash_dir()
logic is used for all cases, remove hard-coded
references to /usr/etc, /usr/share, /usr/local/etc,
and /usr/local/share while providing for
xaccResolveFilePath to actually search the data and
sysconfdir directories used in the build.
(gnc_build_data_path): New function, just a copy of
gnc_build_book_path. Needed for rewrite of
xaccResolveFilePath. (xaccResolveFilePath): Cleaned out
the hard-coded paths and weird file path contruction
functions and rewrote the function to use
gnc_path_get_foo and gnc_build_data_path without all of
the silly indirection. Removed superfluous URI checks
(which are correctly perfomed by xaccResolveURL()).
(MakeHomeDir, xaccPathGenerator, xaccDataPathGenerator)
(xaccUserPathGenerator): Deleted; their functionality is
replaced as noted above. (check_file_return_if_true):
Renamed check_file_return_if_valid, a more descriptive
name.
* Bump glib2 minimum requirement to 2.12. At the same
time, remove all the conditional code and workarounds
that were in the code to cater for glib < 2.12. Note:
this commit will require a rerun of autogen.sh and
configure.
* README referred to a non-existent file So per a
suggestion on IRC, I took the reference out. Patch by
Jesse Weinstein.
* Win32: Update libxslt version, but it needs its own copy
of libxml2.dll. Apparently the gnome-provided libxml2
has the DLL filename libxml2-2.dll, but the binary from
xmlsoft.org still has the filename libxml2.dll.
* Win32: And one more dependency upgrade (causes missing
libxml2.dll complaints otherwise)
* More win32 dependency version updates.
* Update libpng package dependency for win32 to 1.4.0
* Small spelling fixes in the comments
* Re-indentation of source code, next batch. This also
strips trailing whitespaces from lines where they
existed. This re-indentation was done using astyle-1.24
using the following options: astyle --indent=spaces=4
--brackets=break --pad-oper --pad-header Discussed at
http://lists.gnucash.org/pipermail/gnucash-devel/2009-August/026121.html
* Update gnome package versions.
* GDate values weren't being properly fetched from objects
to be saved in a database column if they were fetched as
a gobject property.
* Clean up account column in the lot table to specify that
the guid is an account reference. Simplifies the code a
bit, and makes future use of foreign keys easier.
* If building for WIN32, use webkit_web_frame_print_full()
so that a GtkPrintOperation object with the correct
units can be used to prevent font size problems (see bug
591177). On other platforms, use
webkit_web_frame_print() because some distros seem not
to have webkit_web_frame_print_full() (and also don't
have the font size problem so on those distros, we don't
need to create our own GtkPrintOperation object).
* Cutecash
Some of the GnuCash developers have decided to rewrite the UI
for the cross-platform Qt toolkit from TrollTech. The goal is a
simpler UI which is more powerful and easier to develop. This
project has gotten the name "Cutecash". It uses the same
back-ends and engine as gnucash. Only the UI is different. The
source for the Cutecash UI is in the same tree (and therefore,
the tarballs) as GnuCash, but at this point, no MAC or Win32
builds are being produced.
* Cutecash: Add Commodity wrapper. Use gnc-exp-parser for
numbers. - Allow amounts to be edited. - Let the date
column be handled by the QDate delegate with a QDateEdit
widget
* Cutecash: Enable entering of more cells in register.
Some code cleanup. Add class documentation.
* Enable editing of the "Description" column in the split
list view - WITH UNDO! The Qt Undo framework is almost
like magic. We just have to create a command object
instead of directly manipulating the value, and suddenly
the undo/redo just works. This is fun!
* Cutecash: Add QUndoStack to implement all editing
through the Command pattern and make it undoable.
* Cutecash: Enable closing and re-opening the different
tab views. Also, change many main window slots to make
use of the auto-connection feature because it makes the
slot intention much easier to read. Also, note how we
store the Tab position, title, isEnabled state in
dynamically allocated properties in the Tab widget
itself - this is a rather cool feature of QObject here
(see reallyRemoveTab() and viewOrHideTab()).
* Cutecash: Add Recent-File menu.
* Cutecash: Enable tab moving and other UI features of Qt.
* Cutecash: Add Timespec conversion to QDateTime. Add
display of transaction date in register tabs.
* Cutecash: Display account balance in tree and split
amount in account register.
* Cutecash: Add gnc::Numeric wrapper for gnc_numeric.
* Implement a table widget with the list of the splits of
one account, and open this if an account in the tree is
double-clicked. Date and values/amounts can follow next,
once those types are suitably wrapped into C++ as well.
* Cutecash: Add progress bar during loading the file.
* Implement the account list data model as a
specialization of the account tree model. This is
helpful in order to understand Qt's Model/View
structure, so both (list and tree) are still available.
* Cutecash: Fix guile version number query. Patch by
Herbert Thoma.
* Cutecash: Add a tree view of the accounts.
* Cutecash: Fix CMakeLists for change in guile lookup,
r18846
* Cutecash: Remove QSharedPointer because manual delete is
sufficient. Also, the QSharedPointer cannot be used for
bookkeeping of a C pointer to any gnucash object because
it refuses to work if it doesn't know the actual struct
definition, which in gnucash is always private. The
boost::shared_ptr would work without (by the custom
deleter argument in the constructor), but QSharedPointer
doesn't (the custom deleter is accepted only in addition
to the known storage size, not alternatively), so it is
pointless here.
* Cutecash: Add version check for guile and define
HAVE_GUILE18 if appropriate.
* Adapt cutecash to r18842.
* Cutecash: Fix extern "C" usage: Must not enclose system
includes, supposedly.
* Cutecash: Allow older glib versions as well.
* Cutecash: Add business-core including business-core/xml
into the executable.
* Add copyright notices in cutecash source code files.
* Cutecash: Copy some more icons into the program. Add a
clickable hyperlink.
* More C++/Qt4 frontend work.
* Add example main window in C++ and Qt that links against
gnucash-engine. The example was based on Qt4's
"application" example, but the main window layout is
done through the .ui file already.
* Finish cmake build system for the C++ experiment. To
build this: mkdir build-cpp cd build-cpp cmake ..
make ./src/gnc/cutecash
* C++ experiment: Extend the AccountModel into a table
with name and description. Use QString everywhere as
well.
* C++ experiment: Add first simple model/view widget for
the loaded account list.
* Some more C++ work. Opening an existing file works, even
though nothing is visible so far. The implementation of
a scoped_ptr wrapper around a C object uses the boost
library, though, because writing our own implementation
of a scoped_ptr with custom deletion function is too
non-trivial and I rather re-use the boost::shared_ptr
here instead of making our own mistakes.
Caveats for Testers
Any 2.3.x version might crash unexpectedly at any point during
runtime. If you test some serious work in a 2.3.x release and
are using the XML file backend for data storage, make sure you
hit "Save" after every non-trivial workstep. If you are using
the SQL backend, this is not required as every change is saved
immediately to the database.
The documentation has had screenshots updated, however, many
help texts usually only refers to the 1.8.x series. Everyone is
invited to help improve the documentation; see
http://wiki.gnucash.org/wiki/Development on how to get involved.
How can you help?
Testing: Test it and help us discover all bugs that might show
up in there. Please enter each and every bug into bugzilla.
Translating: The new release comes with some new translation
strings. If you consider contributing a translation, we invite
you to test this release already. A string freeze will be
announced in one of the later 2.3.x releases. Please check
http://wiki.gnucash.org/wiki/Translation_Status for updates on
this.
We would like to encourage people to test this and any further
releases as much as possible and submit bug reports in order
that we can polish GnuCash to be as stable as possible for the
2.4.0 release in a few weeks. Then post any bugs you find to
bugzilla
Getting GnuCash
Source code for GnuCash 2.3.11 can be downloaded from multiple
locations:
* The GnuCash website
* Sourceforge: bzip2, gzip, all files.
* You can also checkout the sources directly from the
subversion repository with this command:
svn co http://svn.gnucash.org/repo/gnucash/trunk gnucash
To install GnuCash, you will need Gnome 2, guile, slib. In
addition you will need swig if compiling from subversion.
Win32 binary
The Gnucash 2.3.11 Win32 setup executable can be downloaded from
Sourceforge as well.
About the Program
GnuCash is a free, open source accounting program released under
the GNU General Public License (GPL) and available for
GNU/Linux, *BSD, Solaris, Mac OSX and Microsoft Windows.
Programming on GnuCash began in 1997, and its first stable
release was in 1998.
More information about the gnucash-devel
mailing list