- cppchecked (1.54)
3 lines of code changed in 3 files:
* more OSX patches
34 lines of code changed in 2 files:
* target 'headerlibs' -> 'header_libs'
- otherwise OSX thinks they are up-to-date (since directory 'HEADERLIBS' exists and OSX is not case-sensitive)
2 lines of code changed in 1 file:
- patch for OSX
2 lines of code changed in 1 file:
- scaling awar (next neighbour search) was defined at wrong place (EDIT4/Aligner/MoreSettings crashed)
1 lines of code changed in 2 files:
* workaround quoting error in ARB_GDEmenus (clang)
11 lines of code changed in 2 files:
- provoke_core_dump via volatile pointer (otherwise clang will skip the code which is crashing on purpose)
2 lines of code changed in 1 file:
* merge from stable_5.0 [8655]
54 lines of code changed in 1 file:
* gb_read_bin
- DRYed code reading keywords
- fixed one wrong return value (hope so)
* DRYed binary-DB id
32 lines of code changed in 1 file:
* fixed gb_put_number / gb_get_number
- numbers above 0xfffffff
* able to write (just did nothing in NDEBUG mode, where that limit may more likely be reached)
* fix reading (returned wrong value)
35 lines of code changed in 2 files:
* added test for gb_put_number / gb_get_number
- showing which values can't be handled
- commented out failing assertion
113 lines of code changed in 3 files:
* moved inlines into own header
* gb_read_number -> gb_get_number
124 lines of code changed in 4 files:
* gb_read_in_long / gb_write_out_long -> gb_read_in_uint32 / gb_write_out_uint32 (that's what it does)
- changed parameter/result to uint32_t
- inlined
- uses fread/fwrite
* gb_read_number
- moved into ad_load.cxx
- inlined
* GBT_check_arb_file
- moved into ad_load.cxx (due to inlining of gb_read_in_uint32)
128 lines of code changed in 7 files:
* GB_is_readablefile no longer returns true for directories
- affects behavior of fileboxes
21 lines of code changed in 2 files:
* allow NULL-arg in
- GB_is_regularfile, GB_is_link, GB_is_readablefile + GB_is_directory
- GB_mode_of_file+GB_mode_of_link returned garbage
* GB_symlink returns reason for failure
* added tests
- showing wrong behavior of GB_is_readablefile for directories
105 lines of code changed in 2 files:
* more IO error handling
75 lines of code changed in 5 files:
* use results from system calls
9 lines of code changed in 5 files:
* fixed syntax error
1 lines of code changed in 1 file:
* removed commented code
0 lines of code changed in 1 file:
* make more obvious what 'Kill' does
* fixed button sizes for aw_question
* use result of system call
9 lines of code changed in 2 files:
* use result from fgets
77 lines of code changed in 6 files:
* fixed warnings
- possible uninitialized variables
- DEBUG-mode variables
- ISLAND_HOPPING/Makefile: disabled warnings (i wont touch that module, except for removal)
90 lines of code changed in 22 files:
* one awar in !ColumnStat_2_gnuplot.cxx clashed with an awar used in !ColumnStat
- used the same name, but had different meaning
* corrected smoothing calculation
- calculation was broken completely since [6261]
- was a bad way to smooth values anyway
* specify smoothing parameter using textfield
69 lines of code changed in 2 files:
- removed some unneeded casts
5 lines of code changed in 1 file:
- dont remove TAGS by clean
1 lines of code changed in 1 file:
- fix aliasing problem
12 lines of code changed in 3 files:
merge from stable_5.0 [8635] (OSX)
13 lines of code changed in 4 files:
merge from e4fix [8443] [8444] [8445] [8604]
* new EDIT4-keys
- fold/unfold group using ENTER key
- ALT+up/down jumps between group consensi
- Ctrl-M marks/unmarks current species or group
* minor help-fixes
* removed "[Ctrl-A]" from aligner menu entry (they do different things)
* use ED4_terminal for owner_of_cursor and ED4_base_position
135 lines of code changed in 14 files:
merge from e4fix [8430] [8431] [8432] [8433] [8434] [8435] [8436] [8437] [8438] [8440] [8441] [8442]
* fixed display concerns
- wrong mark boxes for unmarked species when (slowly) scrolling vertically
- tiny gaps when moving cursor leftwards (best visible if species is selected)
- do not draw all spacer terminals
- rewrote ED4_bracket_terminal::draw (increased bracket-width by 1 pixel)
* marked_species_select and selected_species_mark produced wrong counters since [8625]. fixed
* signature of AW_device::clear_part now same as box_impl
* Rectangle::distinct_from() no longer counts adjacent rectangles as distinct
* AW_screen_area_conversion_mode
- removed FAULTY_OLD_CONVERSION
- added UPPER_LEFT_OUTLINE
* added functions using namespace-AW-Positions
- coordinate transformation
- ED4_base::get_screen_area
* no need for virtual adjust_clipping_rectangle()
- both methods (ED4_base and ED4_terminal) were identical
* get_screen_area -> get_win_area
- decrement size by 1
- use in ED4_bracket_terminal::draw and ED4_base::adjust_clipping_rectangle (eliminating confusing -1 offsets at both places)
* added get_multi_species_manager() to ED4_area_manager and ED4_group_manager
226 lines of code changed in 21 files:
merge from e4fix [8428] [8429] [8605]
* insert/delete now works with running editor (w/o crashing editor)
* do not display helix (or protstruct) information for SAI by default
- added possibility to toggle display for single SAI
- added short helpfile and linked it to other helpfiles concerning secondary structure
* fix edit4-crash when
- setting initial right-click into consensus-terminal
- clicking/dragging somewhere else
135 lines of code changed in 15 files:
merge from e4fix [8418] [8422]
* use !LocallyModify
137 lines of code changed in 6 files:
merge from e4fix [8414] [8415] [8416] [8424]
* cursor was not erased from consensi of folded groups (because ED4_cursor reported itself invisible in that case)
* right click on consensus name now also (un)marks SAI (not only species)
* fix unwanted terminal change
- occurred when 2 edit-windows open and 2 different species selected:
* pressing Ctrl-K in the 1st window moved its cursor to the species selected in the 2nd window
* similar things happened after other actions which changed the folding
- properly protect global awars from unneeded refresh, by telling updateAwars whether the cursor changed its terminal
* replaced one more direct call to ::Show by triggering refresh
54 lines of code changed in 10 files:
merge from e4fix [8409] [8410] [8411] [8412] [8413]
* ED4_list / ED4_list_elem
- moved code to header
- now is a typed template (instead of void*-container)
- replaced list instances with pointers to reduce dependencies
* removed all now unneeded casts from client code
* make explicit that ED4_selection_entry::object always is a ED4_species_name_terminal
* detect useless runtime-casts (to_XXX()) and runtime-type-tests (is_XXX()) by overloading these functions for the destination-type
305 lines of code changed in 16 files:
merge from e4fix [8402] [8403] [8404] [8405] [8406] [8407] [8408]
* SAI/consensus flags
- moved into ED4_species_manager
- changed flags -> ED4_species_type (set via ctor)
- added a bunch of convenience functions to detect "species-type" (normal, SAI or consensus)
* removed private+missing copy-ctors (instead make sure the affected classes are derived from Noncopyable)
* "class public:" -> struct
* moved selection flag into ED4_species_manager
* remove_from_selected / add_to_selected expect an ED4_species_name_terminal as argument
* moved 'selection_info' and 'dragged' into ED4_species_name_terminal
* removed unused function is_primary_selection()
* behavior
- no longer marks SAI
- fixed moving groups (where the user actually moves the consensus name terminal)
628 lines of code changed in 22 files:
merge from e4fix [8390] [8391] [8392] [8395] [8397] [8398] [8399] [8400] [8401]
* fixed cursor handling
- draw cursor as final action
* do not draw together with terminal (owner_of_cursor). If other terminals get drawn later, parts of the cursor will be erased again.
* instead ED4_main_manager always draws the cursor (when everything else was drawn)
- deleting the cursor interfered with refresh flags
* save and restore status of affected terminals (owner_of_cursor and the 2 adjacent terminals)
- is_partly_visible/is_completely_visible now return false if cursor is in folded group
* correctly calculate fix-block-edge when extending a block that is completely out-of-view
* always scroll up/down completely when scrolling first/last terminal into view
* screen_to_sequence: only allow screen_pos==screen_len (not above)
- fixes wrong update range when jumping to end of seq in compressed mode (since [8614])
* Pressing END did not jump to end of alignment, when first gap after last base was folded away
* special_window_refresh: new param whether to handle update requests. Do NOT handle them when deleting the cursor.
* explicitely request refresh of all sequence terminals when remapping changed
* request relayout from ctrl-L
* unclosed TA when no species selected on 'copy current species'
320 lines of code changed in 15 files:
merge from e4fix [8376] [8377] [8378] [8379] [8380] [8386] [8387] [8388] [8389]
* do not save AWARs which have default values
- motivation: edit4-props saved after [8362] did not load with older versions
- pros:
* if the user did not change the default, changing the default in code will propagate to the user
- cons:
* AWARs used by multiple applications will trigger in non-saving application, when the database containing the awar is saved
- only happens once if AWAR was previously stored with default-value
- the most obvious ones were those edit4-search-strings linked to ntree (primer, probe, gene). disarmed them by setting the default to NULL
* always generate definitions of SAI color translation tables (since default no longer saved)
- drawback: when user deletes a default color translation table, it will re-appear after restart
* allow experts to save fast-aligner protection to properties
* deliver errors exported by awar callbacks (when triggered by widget change)
* sequence color mapping
- translate stored old default ('=0') to new default (!'') on awar creation
- removed code talking to AWARs via ARBDB interface - just use AWARs straightforward
* ignore request_refresh_for_sequence_terminals when there is no hierarchy yet (now triggered by seq colors)
* lib/arb_default/edit4.arb
- removed default values (where code defaulted to same value)
- changed some defaults
* aligner (explicit, but empty ref)
* helix setting (stolen from WL)
* lib/arb_default/ntree.arb
- removed all values
* equal to default value
* obsolete entries (former awar names)
* entries that are now stored in main DB
* removed unused AWARS (AWTC_create_rename_awars)
* default for AWAR_TREE now "" everywhere
* removed unused AISC variable
* renamed edit direction related variables/awars
304 lines of code changed in 41 files:
- completes [8316] for NDEBUG mode
1 lines of code changed in 1 file:
merge from e4fix [8340] [8341] [8342] [8343] [8344] [8345] [8346] [8347] [8348] [8349] [8350] [8353] [8354] [8362] [8370] [8371] [8372] [8373] [8374] [8375]
* changed refresh behavior
- refresh_all_windows
* call via timer (every 2 sec)
* replaced by refresh-requests where needed
* added ED4_trigger_instant_refresh
- set_refresh -> request_refresh
- removed duplicated refresh-request in ED4_consensus_definition_changed
- replaced ED4_refresh_window by ED4_request_full_refresh
* some callers now request fewer refreshes
* all callers now cause refresh of ALL windows
- ED4_cursor::!ShowCursor no longer updates global AWARs when used for display only
- refresh_window does not handle updates (it is only used to quickly draw the cursor)
* renamed into special_window_refresh() to disencourage usage
- removed implicit clearing of refresh flags previously done when refresh propagated down through windows, areas and hierarchy. Instead refresh_all_windows explicitely calls clear_refresh when done.
* cleaned up resize requests
* removed extra sync in ED4_resize_cb
* removed special handling of 1st call to ED4_expose_cb (no idea why, but this fixes an ancient bug: "save props in hide-all-gaps mode")
- no longer abuse ED4_expose_cb
* trigger relayout
- when helix/seq spacing is changed
* ED4_WinContext
- missing when NDS changed
- expect one in ED4_root::special_window_refresh
- provide from ED4_cursor::delete_cursor
* EDIT4 scrolling
- scrollbar code refactored
- SLIDER_OFFSET was missing in calculation of lower/right folding lines (fixed)
- picture size announced by ED4_window::update_scrolled_rectangle was off by 1
- fixed scroll/resize bugs introduced in [8618]
* added debug code
- print backtrace on negative folding line dimension
- print mismatch if slider postions and folding line dimensions are out of sync
* sequence color mapping
- fixed broken window layout
- increased number of sets to 8
* added symbolic ambiguity set as default
- hide default value '=0'
* group name generation (index-safety)
* execute callback-guards and post-useraction-callback whenever an awar change is directly triggered by a widget
* moved (ARB_)strNULLcmp to global header
* doc/fix unexpectable behavior (NULL is bigger that any text) with a test
* dont use magics for system folder
931 lines of code changed in 43 files:
merge from e4fix [8303] [8304] [8305] [8306] [8307] [8330] [8331] [8332] [8333] [8334] [8335] [8336] [8337] [8338] [8339]
* moved variables handling selected block into new class ED4_block
* removed ED4_species_container_changed_cb
- removed GB_search_last_son (only used there)
* fixed '(re)load config'
* ED4_WinContext missing
* in create species functions
* in ED4_remote_set_cursor_cb
* when changing HELIX-inputfield
* NULL-deref in ED4_host::announce_current_species (if no species selected yet)
* DRYed get_area_level
* get_upper_lower_cursor_pos
- no longer a member of ED4_cursor
- replaced callback+cb-arg-param by terminal predicate
- loop backwards through hierarchy when moving cursor up
- DRYed duplicated code in loop
* calc_world_coords now const
* when moving cursor vertically, scroll display if target terminal is only partly visible
* fixed scroll_into_view
* moved visibility checks from ED4_base to ED4_window
* set_properties -> set_property (added inverse function clr_property)
* ED4_manager::un/fold_group moved to ED4_bracket_terminal::un/fold
* do group-size calculation via update flags (first request; recalc during refresh)
* when creating new group: correctly calc group size and selection status for source group(s)
* fixed a buffer-overflow in group_species
* display ORF terminals at cursor of MOST RECENT window (hack)
655 lines of code changed in 22 files:
merge from e4fix [8239] [8275] [8276] [8277] [8286] [8288] [8289] [8290] [8293] [8295] [8296] [8298] [8299]
* define window-contexts by instanciating context-objects
- ED4_LocalWinContext (used where calls are triggered via a window)
- ED4_MostRecentWinContext (uses "last used window"; used when window is not known)
- dump backtrace on missing context
* window-context changes
- ED4_search_cb (ensure search is performed in correct window)
- ED4_set_iupac
* fixed ED4_compression_changed_cb (now updates all windows when compression changes)
* made several functions context independent
- ED4_calc_terminal_extentions uses any device (that of first window)
- ED4_expose_recalculations no longer does update_scrolled_rectangle (now performed by caller; caller knows the context)
* added special class ED4_WinContextFree (denies access to global context for derived classes)
* each cursor remembers the window where its located
* declared ED4_window and ED4_cursor as ED4_WinContextFree (the window is their explicit context)
- replaced invalid references to global context by uses of explicit context (get_device() + window())
* cleaned ED4-object-deletion (what a mess)
- rewrote ED4_terminal::kill_object using Delete()
- regenerate group-ids during refresh (after deletion took place)
- DRYed delete mechanisms
- announce_deletion down to ED4_base_position (to invalidate its cache)
* 'Get' selected species failed to
- update bases-tables of parent-managers (that was one cause for folding errors!)
- regenerate group-ids
* syncronize mode in all edit windows
* use position/size from awars when creating a 2nd window with identical ID
* removed some useless virtualization
* denglish (actual->current)
691 lines of code changed in 27 files:
merge from e4fix [8227] [8228] [8230] [8231] [8232] [8236] [8237] [8238] [8329]
* ED4_scrolled_rectangle
- made all members private
- use AW-objects
- ensure folding line dimensions were calculated before get_window_rect is called
* ED4_foldable
- extracted from ED4_window (enables separate tests)
- added tests
- using multiple folding lines is broken by design
- removed link-param from insert_folding_line (always passed as NULL)
- removed 'length' (all have infinite length now)
* ED4_folding_line (code from hell)
- added ctor/dtor (fixing leak)
- fix tests (fixed world_to_win_coords())
- added tests for world positions inside, nearby and at edges of folded range
- made all members private
- reduced to needed minimum: position, dimension, next
- now does world<->win transformation
* other changes
- avoid duplicated background-erasure in calc_bounding_box()
* only call get_upper_lower_cursor_pos with screen coordinates when cursor is in top-area
451 lines of code changed in 14 files:
merge from e4fix [8191] [8192] [8193] [8194] [8195] [8200] [8201] [8202]
* removed unused default parameter 'only_text' from ED4_terminal::draw
* removed implementation of pure member functions (draw,Show)
* ED4_root
- replaced initialization via memset with explicit member-initialization-list
- added default-ctors for those member w/o any
* introduced ED4_WinContext
- (currently only stub replacing tmp_aww etc)
- wrapped access to current device/aww/ed4w
* added AWT_install_postcb_cb
- installs a callback which is called after most user interactions (when action-cb finishes)
- track most-recently-used edit-window
* world_to_win_coords / win_to_world_coords
- removed useless parameter AW_window
- moved into ED4_window
- refactored
- bugfix
401 lines of code changed in 27 files:
merge from e4fix [8190] [8196] [8198] [8197] [8199] [8221] [8226]
* callallcallbacks
- allow repeated calls for all orders
- avoid infinite descent into recursive filters
- avoid config managers
* fixes via callallcallbacks
- some un-/over-handled errors
- missing transaction
* make
- optionally keep preprocesser output
* tests
- only explain broken/wanted expectations if warning was shown
- added TEST_EXPECT!__BROKENIF
* spelling
- intervall -> interval
117 lines of code changed in 22 files:
* source-list-fix for [8614]
1 lines of code changed in 1 file:
merge from e4fix [8168] [8169] [8170] [8171] [8178] [8180] [8181] [8182] [8183] [8184] [8185] [8187] [8188] [8189]
* merge similar concepts !UpdateRange and !TargetRange into !PosRange
- moved to CORE
- added !ExplicitRange
- added PosRange::contains()
- able to intersect
- use in
* !AlignParams
* readCompactedSequence()
* !UnalignedBases + !UnalignedBasesList -> !LooseBases
* to handle selection (aka "block") in edit4
* for update interval (draw functions)
* build_probe_match_string
* added !AliChange
- holds sequence before and after alignment
* DRYed code of calc_intervall_displayed_in_rectangle vs calc_update_intervall
* cache sequence length in ED4_orf_terminal
* added test for !UnalignedBasesList (documenting wrong behavior)
* renamed all *.stamp to stamp.*
905 lines of code changed in 35 files:
* avoid upcoming nameclash caused by defining 'contains' in unit-tester
- use less common words:
* contains -> does_contain
* equals -> is_equal_to
* differs -> does_differ_from
51 lines of code changed in 5 files:
merge from e4fix [8172] [8176]
* tests execution order
14 lines of code changed in 6 files:
merge from e4fix [8160] [8161] [8162] [8163] [8164] [8165] [8166] [8167]
* renames
- ED4_species_pointer (get/set methods)
* coverage
- reports now controlled via Makefile.setup.local
- fixed failure propagation from gcov2msg.*
* fixed
- return type of ED4_create_faligner_window
- several free/delete mismatches
* regression test for ED4_char_table (build on [8159])
- loosened dependency of ED4_char_table from global ED4_ROOT (to get test working)
* consensus-update-methods
- pass !UpdateRange as object
* build_consensus_string
- removed default-params (instead use diff functions flavors)
- no longer uses uncommon offsets
- no longer propagate "no changes" upwards through consensi
567 lines of code changed in 27 files:
merge from e4fix [8156] [8157] [8158] [8159]
* changed condition result to bool
* bugfixes (RUI, free/delete, SEGV)
166 lines of code changed in 15 files:
merge from e4fix [8145] [8146] [8147] [8148] [8149] [8150] [8152] [8154] [8155]
* removed useless empty dtors
* directly test !BasePosition (removed test fakes)
* build
- select source files using wildcards
- removed ED4_secedit.cxx (failed to delete in merge [6865])
* introduced ED4_objspec
- replaces dull struct ED4_object_specification
- instanciate them via ctor
* add all ED4_objspecs to ED4_objspec_registry (which does several validity checks)
- cant add objects to wrong parents
- expect an accepting parent-object-spec for each object-spec
- removed justification (unused)
- maintain possible_descendants
* announce actually added members to ED4_objspec
* changes to object-specs:
- fixed allowed_children for ED4_main_manager, ED4_device_manager, ED4_species_manager and ED4_sequence_manager
- group_manager_spec no longer allows adding ED4_device_manager (root_group_manager_spec does)
640 lines of code changed in 21 files:
merge from e4fix [8151] [8153]
* failed attempt to use stabs+ instead of default dwarf
- works better for debugging (with gdb)
- fails completely with valgrind
- left it there in comments
9 lines of code changed in 5 files:
merge from e4fix [8135] [8136] [8137] [8138] [8139] [8140] [8141] [8142] [8143] [8144] [8222]
(this revives the reverted patches [8129] [8130] [8131] [8132]; see [8133])
* fixes
- some free/delete mismatches
- wrong definition of ORF objects (Level was no bit value)
- amino consensus (failed for columns only containing 'C')
* rename
- AA_sequence_term -> orf_term
- ED4_sequence_terminal_basic -> ED4_abstract_sequence_terminal
* cleaned up hierarchy dumps
* tweaked is_terminal()/to_terminal()
566 lines of code changed in 193 files:
* missing source def [8603]
* show future errors as warnings during normal build
41 lines of code changed in 2 files:
* postprocess macro code and remove duplicated awar-changes
255 lines of code changed in 6 files:
* added tests for reading different LFs
65 lines of code changed in 7 files:
* added class !FileContent
117 lines of code changed in 6 files:
* callback introduced in [8597] was not removed when an input widget triggered, but value hasn't changed
14 lines of code changed in 1 file:
* options to (execute and then) expand existing macros
* rewrote macro help
172 lines of code changed in 7 files:
* added macro that records an ARB session via
{{{
arb_ntree --execute _dolog database.arb
}}}
30 lines of code changed in 1 file:
* ensure macro code is written even if ARB crashes as direct consequence of the action (e.g. because an awar or button callback crashes)
* as well allows to record "quitting ARB"
23 lines of code changed in 3 files:
- DRYed
11 lines of code changed in 1 file:
- avoid writing to macro file after an error occurred
- fix strange error message when no macro name given
19 lines of code changed in 3 files:
* fix ARB crash when recording unrecordable actions
- e.g.
* SpeciesInfo/<select a db field>
* !UserMask / Edit button (no crash here - just not recorded)
* instead warn about
* set stop_action_name to correct value (dont warn that pressing the stop button was not recorded.. :)
27 lines of code changed in 4 files:
* Awar type AW_POINTER is only used for pointers to GBDATA
- restrict usage to this fact
Note: AW_POINTER was introduced in [6146:6148]
26 lines of code changed in 9 files:
* use strict + warnings
* declare variables
* use await_error (get_error fails)
8 lines of code changed in 1 file:
* moved macro code into own class and module
185 lines of code changed in 6 files:
* missing macro id
1 lines of code changed in 1 file:
* macro execution
- when a macro terminates (closes DB), it informs the caller via DB about that
- AW_root::execute_macro(): new callback param; will be called after macro termination
- when running a macro from GUI, it's termination callback will reset the currently selected macro
* just cosmetic, but proofs concept works
* only annoying when calling a macro which calls another macro (previously that other macro was selected after execution)
* GB_close
- run close callbacks before disconnecting from server (-> close callbacks now can transact with DB)
168 lines of code changed in 12 files:
- moved SL/FILE_BUFFER into CORE
280 lines of code changed in 18 files:
* create consensus tree from gui
179 lines of code changed in 10 files:
- instead of terminating arb_ntree, print error message about accessing unknown variables (awars)
27 lines of code changed in 1 file:
* moved code from arb_consensus_tree -> class !ConsensusTreeBuilder
86 lines of code changed in 2 files:
* fixed dependency->target resolution for object-files (i.e. for small binaries build upon one source files, like done in TOOLS/)
- related to [8574]
23 lines of code changed in 1 file:
* rename
7 lines of code changed in 1 file:
* added generic subset-selection (refactored from yadhus code in 'concat alignments')
- uses an existing string-selecttion-list (content does not matter)
* creates a second, empty list
* creates buttons for subset selection and sorting
* concat alignments
- use generic subset-selection for alignments
- bugfix: generated alignment was too short (separators were ignored), always needed to format it after generation
* added global defines for default-selection
358 lines of code changed in 14 files:
* remove unused member AW_at::correct_for_at_string
* also store y_for_next_button in AW_at_auto (fix unwanted oversize of some windows introduced by [8579])
7 lines of code changed in 3 files:
* function to create and bind re-order buttons
- use in tree-admin
- use in ali-concat
115 lines of code changed in 9 files:
* possibility to store/restore auto_space/auto_increment
50 lines of code changed in 2 files:
* path fix related to [8574]
9 lines of code changed in 1 file:
* generic alignment selection list now able to reconfigure on ali-type
- use in NT_concatenate (and DRYed some code)
61 lines of code changed in 4 files:
- skip generated file
1 lines of code changed in 1 file:
* inlined PT_BASES -> char conversion
* removed unused length-parameter from PT_base_2_string, reverse_probe and complement_probe
- saves several calls to strlen
* added debug code to dump postree to file on load (inactive, see PTM_DUMP_PTREE)
* changed existing dump-code
- dumps to file instead of stdout
- tighter output for species+pos ('spec@abs>rel')
78 lines of code changed in 7 files:
* [8567] failed on WCs where shared libs were up-to-date, but not yet present in $ARBHOME/lib
- link libs directly in $ARBHOME/lib
* no longer create 'SHARED.a' in shared-library-dirs
* unit tests caused build of 'SHARED.so' (instead of 'libSHARED.so'). Every library existed twice.
- changed test-target-names for shared libaries (e.g. ARBDB/ARBDB.test -> ARBDB/libARBDB.test)
- added failing stopper targets to main-Makefile and shared-lib-Makefiles
* changed all dependencies to shared libs
* fixed dependency scripts to perform correct translations from lib/libUNIT.so <-> UNIT/UNIT.(dummy|a). omg :/
* updated predefined RESTRICT_LIB entries in UNIT_TESTER/Makefile.setup.template
- check whether RESTRICT_LIB uses obsolete test-names (e.g. ARBDB instead of libARBDB) and fail if.
197 lines of code changed in 55 files:
- increased unit-test timeouts
4 lines of code changed in 1 file:
- test failed if lib/pts did not exist. now only warn
13 lines of code changed in 1 file:
- generated dep.alltargets differed between normal 'make' and make called via 'remake_after_change.pl'. fixed by exporting ARCHS_PT_SERVER_LINK to submakefiles
2 lines of code changed in 2 files:
* when test_environment failed to setup/cleanup (due to crash in itself or any command called to perform the setup/cleanup)
- the error did not propagate to test-suite log [fixed]
- the flag used to synchronize concurrent setup/cleanup attempts was not removed (causing the next test run to fail with timeout). Now they get listed, then removed and a warning is logged.
* left two test-crashes (disabled; in case you'd like to try it)
19 lines of code changed in 3 files:
- updated libdepends changed by [8557]
3 lines of code changed in 2 files:
- exclude LIBLINK from svn-ignore
- ignore some optional dirs for 'make save'
3 lines of code changed in 1 file:
remove LIBLINK
15 lines of code changed in 13 files:
really fix "make clean; make build"
I'm tempted to claim that there is a bug in make. The libs: target
does not work as expected. During a normal build, the target is run,
but the files aren't copied. "make libs" works as expected, though...
=> Reverting r8565 and adding LIBLINK/ to LD_LIBRARY_PATH instead
-- a bit brute, but works.
2 lines of code changed in 1 file:
Have HELP_SOURCE target require libs in lib/
(fixes "make clean; make build" fails)
It doesn't suffice for link_db to have been built as the libraries required
by arb_help2xml need to be in the LD_LIBRARY_PATH (ie $ARBHOME/lib).
(superseded by [8566])
1 lines of code changed in 1 file:
* arrow-buttons
- all are standard size now (17x17)
- horizontal arrows use same style as vertical ones
* cleaned layout of
- alignment concat window
- tree admin window
* fixed error message on gfx-load-failure
193 lines of code changed in 14 files:
* merge tool (trees)
- transfer all, existing or missing trees
- optionally overwrite trees
159 lines of code changed in 3 files:
* added basic tests for !StrArray
- fixed bug ({{{array[0]}}} crashed on empty array)
35 lines of code changed in 2 files:
* MERGE
- renamed AWARs and DB names (using 'src' and 'dst')
* no saved awar-paths changed - so user-properties should be unaffected
* replaced magic awar names by defines
* strictly separate between DB-related and DB-independent awars
* dynamic awar names now generated via central function
- DRYed some code existing for both DBs
531 lines of code changed in 16 files:
* merge tool (trees)
- transfer trees with order: trees get inserted behind the tree selected in the dest-tree-list
* after a tree has been transferred, step to next tree in both lists
- => repeatedly clicking 'Transfer tree' copies a sequence of trees
- fixed initial value for tree awars
- extracted plain transfer-code into separate function
74 lines of code changed in 2 files:
* fix bug in GBT_get_next_tree: never return input tree
- caused tree to stay selected when deleting last tree
* added several tests for 0-2 existing trees
* when no tree selected, dont complain about illegal chars in tree name (tree_?????) on copy/rename attempt
* test for GBT_size_of_tree was not broken
90 lines of code changed in 2 files:
* global define for "no tree selected"
40 lines of code changed in 14 files:
* moved some tree admin functions into SL/TREE_ADMIN
- use them from 'Tree admin' and from 'Merge trees'
407 lines of code changed in 15 files:
* implement the 3 disfunctional tree-order buttons in 'Tree admin'
- see #155
* renamed and changed button gfx
68 lines of code changed in 5 files:
* tree-functions
- rewrote/changed search functions (especially positional search)
- added GBT_copy_tree and GBT_rename_tree (code was in NTREE/ad_trees.cxx before)
- name functions now all return 'const char *'
* added/adjusted tests
* fix broken tree-order (duplicate indices) when detected
* use new tree functions from 'Tree admin'
443 lines of code changed in 9 files:
* fixed GB_number_of_subentries (did report wrong number when in transaction mode - i.e. nearly always)
- now always counts
22 lines of code changed in 3 files:
* doc bug in GB_number_of_subentries
30 lines of code changed in 1 file:
* support an order on trees
- generated on-the-fly if missing (using old database order)
- added GBT_move_tree() and its tests
168 lines of code changed in 6 files:
* automatically popdown key selection prompter in NDS window
17 lines of code changed in 3 files:
* fixed locals in GB_resort_data_base
20 lines of code changed in 1 file:
* YAMWF (yet another modal window fix)
- use activate instead of show (this fixes problems with KDE where several windows popped up in the background)
* from AW_POPUP callbacks
* from show_modal()
- windows using stored positions
* when show was triggered after window has been moved, it always jumped back to the previously stored position.
* now stores (changed) position on show (if window is already shown)
- windows centering on mouse-cursor
* always center to cursor (not only when cursor is outside window)
- correct window position against screen size
32 lines of code changed in 1 file:
- gb_search: another error message (key='..sub')
4 lines of code changed in 1 file:
* fixed gb_search
- better error messages (especially for invalid keys)
- no longer accepts NULL-key
- fixed several crashes (e.g. when key contained '*')
- separated code handling allowed separators ("/", ".." and "->")
- trailing slash allowed for containers (not for terminals)
- single slash allowed (returns main-entry)
- several invalid characters were silently interpreted like '/' and created containers. now they report an error
132 lines of code changed in 3 files:
* rename
4 lines of code changed in 2 files:
* changed some tree-load error messages
14 lines of code changed in 3 files:
* CLI-tests
- report output in case of error (including backtrace if any CL-tool crashes)
- added !__BROKEN flavors
49 lines of code changed in 2 files:
* tweaked readabilty of failing error-tests
- for functions reporting errors via GB_export_error
- for functions returning errors (always have to specify part of error message)
* typo ('exacly')
79 lines of code changed in 4 files: