1.19.2: 2010-11-23 Murray Cumming Fix the tests build during distcheck * Makefile_glom.am: Correct eggspreadtablemm.c to eggspreadtablemm.cc. * Makefile.am: * Makefile_tests.am: Distribute the albums.csv test file and put the path to its directory in a define. * tests/import/test_parsing.cc: Use the define here, instead of using get_current_directory, which is not enough when builddir!=srcdir. 2010-11-17 Javier Jardón Use upstream gettext instead Glib one * configure.ac: Do not use AM_GLIB_GNU_GETTEXT() because it is no longer maintained/recommended. This fixes bug #631367 2010-11-17 Murray Cumming Import tests: Uncomment some tests and make them test what we really want. * tests/import/test_parsing.cc: Check that we do _not_ require ending newlines and check that we do _not_ require quotes around items. 2010-11-15 Murray Cumming Import tests: Disable more tests that do not seem useful. * tests/import/test_parsing.cc: Comment out the non-comma separators test and the ignoring-space-between-quoted-text test, because they don't seem to correspond to any specification. 2010-11-15 Murray Cumming Import tests: Disable tests that check for incorrect or useless behaviour. * tests/import/test_parsing.cc: Comment out the test to ensure that we - Ignore rows with no ending newline. - Ignore field values that are not quoted. because we should not do that. Also add comments explaining the regular expressions. 2010-11-14 Murray Cumming Re-enable the import tests and partially deal with their race condition. * Makefile_tests.am: Re-enable the import tests. * glom/import_csv/csv_parser.[h|cc]: Add ensure_idle_handler_connection(), moving the idle signal handler connection there. Also call it from the beginning and end of on_file_read(), in case it has been disconnected too early. Use a priority less than PRIORITY_DEFAULT_IDLE, so that IO is likely to happen more often than the idle callback. 2010-11-12 Murray Cumming Import tests: Slight code-style change. * tests/import/utils.[h|cc]: Remove the MainLoop typedef because it just obscures things. 2010-11-12 Murray Cumming CSV Import: Actually preview and import the field values. * glom/import_csv/csv_parser.cc: advance_field(): Remove an excess continue, so that we actually add the character to the field's text. This error was probably introduced while removing the non-exceptions ifdefs. This fixed bug #625693 (maximiliano). We really must fix those import unit tests so that something like this does not happen again. 2010-11-11 Murray Cumming List view: Make the retry option actually work afer entering invalid data. * glom/mode_data/db_adddel/db_adddel.[h|cc]: on_treeview_cell_edited(): If the user chooses to retry the edit, remember what he entered, and restart the editing (in an idle handler, to avoid confusing the treeview), with that text. This fixes bug #167818 2010-11-10 Murray Cumming Update EggSpreadTable from libegg. * glom/utility_widgets/eggspreadtable/eggspreadtable.c: Copy from libegg/libegg/spreadtable/ again, to use Tristan's corrections for the forall() vfunc. 2010-11-08 Murray Cumming Call xmlCleanupParser() because libxml++ does not anywmore. * glom/main.cc: main(): Call xmlCleanupParser() to clean up libxml to help valgrind. 2010-11-08 Murray Cumming > FlowTableWithFields: Fix some widget expansion behaviour. * glom/mode_data/datawidget/label.cc: Call Gtk::Label::set_line_wrap() to prevent the label from sometimes making the window insanely wide by demainding too much width for itself. * glom/mode_data/flowtablewithfields.cc: add_field_at_position(): Tell the field's widget to expand, so it uses more space if the user expands the window, instead of just expanding the gap between columns. Note that the old non-EggSpreadTable code only expanded the right-most column, so exanding both is a real improvement. 2010-11-08 Murray Cumming EggSpreadTable: forall(): Avoid use of invalid GList items. * glom/utility_widgets/eggspreadtable/eggspreadtable.c (egg_spread_table_forall): Use (and later free) a copy of the list, because the callback could call egg_spread_table_remove() which would change the list, causing us to use a list->next that is no longer valid. This fixes a problem found by valgrind. I don't know why other Gtk containers don't need to do this, though their code seems to do the same thing. 2010-11-08 Murray Cumming Added test_flowtablewithfields. * Makefile_tests.am: * glom/mode_data/test_flowtablewithfields.cc: Added the beginnings of a test of FlowTableWithFields. 2010-11-05 Murray Cumming Use EggSpreadTable instead of GtkSpreadTable (not added to GTK+). * glom/utility_widgets/eggspreadtable/eggspreadtable.[h|cc]: * glom/utility_widgets/eggspreadtable/testspreadtable.c * glom/utility_widgets/eggspreadtablemm/eggspreadtablemm.[h|cc]: * glom/utility_widgets/eggspreadtablemm/private/eggspreadtablemm_p. h: New files. * glom/Makefile_glom.am * glom/Makefile_tests.am: Mention the new files. * glom/utility_widgets/flowtable.[h|cc]: Use Egg::SpreadTable instead of Gtk::SpreadTable. 2010-11-04 Murray Cumming Remove the gconfmm dependency, because we don't use it. * configure.ac: Do not check for gconfmm. * glom/bakery/app_withdoc.cc: Do not include gconfmm.h. 1.19.1: