LablGTK is an Objective Caml interface to GTK+. This version, LablGTK2, has
support for GTK+ from 2.0 to 2.6.
It uses the rich type system of Objective Caml 3 to provide a strongly typed,
yet very comfortable, object-oriented interface to GTK+. This is not that
easy if you know the dynamic typing approach taken by GTK+.
* Fixed several reference-counting problems in rarely-used methods.
(Murray Cumming)
* Some MSVC++ .Net 2003 fixes. However, a compiler bug stops this from
working properly. (Timothy Shead, Cedric Gustin)
* Some SUN Forte fixes. (Murray Cumming)
* Documentation: book/menus_and_toolbars/: Show how to use a
non-stock icon. (Fredrik Arnerup)
Changes:
Version 1.1
* The intensity of rain and storm is configurable now. New options:
-rain <0-10> and -storm <0-10> (the old yes-or-no variants are deprecated).
* The viscosity of the fluid is no longer a constant (without any speed
loss). New option: -viscosity <1-5>
* The rain feature no longer prevents from the idle mode: just obscure by
other windows, but don't give -nww.
* Bugfix: On 64-bit x86, xdesktopwaves now even compiles with gcc 3.4.
* Bugfix: Fixed arithmetic exception on systems with a small RAND_MAX.
* Bugfix: If -stippled, clear window at initialization.
* Optimization/Bugfix: The color palette is reduced to have only different
colors. This can spare about 10% of the rectangles sent to the server,
if the display depth is 16-bit.
* Optimization: Treatment of minimum range of out-of-date canvas rows. This
spares up to 50% client CPU cycles when there are just very few waves
(shortly before entering idle mode).
This little thingy allows you to start X with NumLock turned on (which is
a feature that a lot of people seem to miss and nobody really knew how to
achieve this).
Changes:
5.14.1
Fixed PAM goof.
5.14
fadeplot,laser: do not clear so much with erase modes
(void *) for (caddr_t *) in argtype to get rid of most gcc warnings.
g++ cleanups for xglock, xmlock, flow, life, marqee, apollonian, pacman
configure cleanups.
Goof for random placement of labeling for life and life3d i.e.
the S could burn in on those old monitors.
flow mode improvements added thanks to Tim Auckland
Replace 2D rendering of Periodic Attractors with a 3D
'interrupted' rendering. Replace "-/+allow2d" with "-/+periodic"
Replace all ODE formulae with completely generic forms.
Add '-search' option to perform background high-speed discovery
for completely new attractors without impacting rendering
performance.
Use gaussian distribution for initial point positions and for
parameter search.
Add "+dbuf" option to allow Double-Buffering to be turned off on
slow X servers.
Remove redundant '-zoom' option. Now automatically zooms if both
rotation and riding are permitted.
Replace dynamic bounding box with static one pre-calculated
during discovery phase.
Simplify and fix bounding box clipping code. Should now be safe
to run without double buffer on all XFree86 servers if desired.
Sync'd up with xscreensaver version
Orbital camera (-/+rotate)
specify following bee (-/+ride)
zoom between bee and orbit (-/+zoom)
Bounding Box (-/+box)
disable 2D Poincare' section flows (-/+allow2d).
Also added basic double-buffering to avoid XFree86 'debris' problems
when lines go out of bounds.
A few more fixes from Eric Lassauge <lassauge AT users.sourceforge.net>.
gtk2/gtk detection in configure, automatically does a make for xglock
xglock now uses GTK2
genlauncher fix for noof (used by xglock).
life1d repeating patterns have a better chance of being caught.
compile dependencies for ImageMagick for VMS thanks to Jouk Jansen.
NICE_ONLY build option added. --enable-nice-only or --enable-blank-only
nice modes also reevaluated thanks to Tom Schmidt.
config.sub and config.guess updated from GNU.
Changed PAM to USE_PAM to be consistent with imake.
loop is more random when run as a single mode. -langton option added
if one just wants Langton's Loops.
atunnels resync with xsreensaver for new textures
thanks to Eric Lassauge <lassauge AT users.sourceforge.net>.
Changes:
- Added Bill and Big Penguins themes
- New ToonCheckBlocked() function to avoid changing toons dying immediately
- Added Brian Chu's superpenguin that can face in both directions
- Action toons that move left or right no longer get stuck at walls
- Use `--squish' to kill the toons with the mouse
- Code uses _() around strings to allow future i10n
- Desktops only: redraw icons by sending expose events to root window
- Enabled multiple themes to be selected
- Use the `--all' option to run all available themes simultaneously
* Fixed possible deadlock in Tcl_Finalize
* Fixed thread IDs on 64-bit systems
* Fixed [scan %ld] handling LP64 systems
* Fixed [string map $samevar $samevar] crash
* Fixed broken [namespace forget] logic
* Fixed [namespace import] cycle prevention
* Improved control over int <-> wideInt conversion
* Fixed buffer overflow when there is a syntax error in [subst]
* Fixed [file executable] result for case insensitivity on Windows
* Fixed [expr {NaN == NaN}] result
* Make [tcl_findLibrary] search $::auto_path as well
* Fixed crash in [fconfigure -ttycontrol] in threaded build
* [TIP 222] implementation of [wm attributes -alpha] on Windows
* [TIP 159] implementation of [wm iconphoto]
* Make Tk [load]-able on OS X/Aqua
* Handle color/screen resolution changes on Windows
* Fixed [wm resizable] for Solaris/CDE
* Handle Windows native dialog double clicks that fall through
* Fixed [place forget] memory leak
* Recompute menu geometry on named font size change
* Fixed X/IME IC focus after creation
* Improved multiple display Tk dialog window placement
* Fixed compound buttons for OS X/Aqua
* Documentation improvements and test suite expansion
Intermediate release 4.50.99
-------------------
* Changed the configuration file for the build. The auto(conf|make) people
changed their API so we changed our build.
* The string-to-pixel-table converter had gone missing; fixed.
* Several resources were not getting treated in SetValues and GetValues.
Some must have been disabled for a while, several others died because
moving them into the per-cell structure had the side-effect that they
weren't any regular widget resources any more.
I'm afraid this was oversight on my behalf during the per-cell conversion.
* Changed the functionality of drawCellCallback slightly: we're now initializing
the callback structure with the cell contents. If the callback doesn't
alter anything, something will still end up getting displayed.
* New work from Tobias Oed :
- setlabel.patch corrects the buggy logic of XbaeSetRowLabel (and ColumnLabel).
- SCROLLBAR_TOP.patch introduces two simple macros to make things a bit more readable
- clips.patch redraws the left/right/bottom/top clips in the same way the clipchild is
redrawn. IMHO it makes things cleaner and avoids redrawing the partially
fixed cells twice in some cases. Also fixes bugs when deleting a row/col
makes the matrix scroll.
- ClearCell.patch fixes a small bug I introduced in GetCellClip.patch (one of the draw
examples crashes).
- deadspace-clipmask.patch redraws the dead space similarly to the labels (they are
both drawn on the matrix). This makes things cleaner as now everything gets redrawn
the way clipped cells are or the way labels are. It also eliminates the need to redraw
the last non fixed row/columns to get the dead space to be drawn. It kills
calcVertfill/calcHorzfill and removes a bunch of unnecessary
- xbaeSetClipMask calls
and simplifies that function. It gets the shadow to span the dead space (when there are
trailing attached rows/columns) and one of XmGRID_ROW_SHADOW etc are set.
It simplifies the way a highlights are drawn and deprecates HighLightOther but these
do not span the dead space yet.
* Protect against passing gadgets as cell widgets for now, because we don't have the
right support for them yet.
* More work from Tobias :
- refresh.patch adds stuff I missed in GetCellClip.patch that leads to this bug: start
choice, add some (trailing)fixed columns, set cell margin heigtht to 10, scroll to the
bottom of the matrix and then reduce the margin height. The fixed column don't get
updated right.
- left_column.patch. Do the same thing with the above patch applied. Everything gets
updated correctly now, but the scrolling is somewhat erratic. This patch fixes this by
making left_column/top_row 'fake' resources (like cells now is). It relies on the assumption
that SetValuesHook gets called after SetValues. If someone has a better X doc than me, let
me know if this is safe.
- filletc.patch gets the fill(s)/trailing attached to work under all combinations of these
options and shadow types according to spec (these are wierd). The approach I started
to take in deadspace-clipmask.patch was a hopless nightmare and this is much simpler. It makes
the fill part of the cell that preced it. To accomodate this, the size of the clips is changed.
It removes the more obscure macros (Sir Silly Macro's) and introduces new ones
VISIBLE_FIXED_*_{HEIGHT|WIDTH}. The new version of exaples/add lets you try the stuff out.
- renamemacro.patch does the renaming I suggested.
- rowlabelwidth.patch fixes the calculation of the row label width as it included
cell_shadow_thickness twice when the labels were buttons.
- labelClip.patch - This patch puts the scrollable parts of labels in their own clips. It adds
two clips and two scrollMgrs but on the other hand kills 3 gcs and the matrix scrollMgr.
Now everything that scrolls is treated the same way and we have as many scrollMgrs as clips.
Time to put them in the clip.
- This patch fixes a few more problems due to the per_cell change of arcad: Set/GetValuesHook
were missing some items and SetValuesHook wasn't updating the display.
I also changes the semantics of per_cell[][].shadow_type a bit: a value of 0 now
means to use the default matrix.cell_shadow_type. Unfortunately there is no
XmUNSPECIFIED_SHODOW so this may not work on all motif flaviours. On
the other hand, the values currently used come from two different motif enums
so it's either already broken or safe.This gets that part of examples/choice to work again.
I modified examples/choice to get the Arm option to work again (You can't just
fiddle with what GetValues returned for cellBackgrounds and cellShadowTypes
and do a redisplay, you need to SetValues what you changed). I also fixed the
memory leaks that every other third party app will suffer after a GetValues of
these resources. That probably needs to be documented somewhere.
- fifteen.patch gets the shadow to follow the empty cell again and fixes some memory leaks
(same thing as my latest changes to choice.c, all other examples are OK)
- labelClipfix.patch fixes a bug I introduced in labelClip.patch (very apparent when scrollbars
are TOP and/or LEFT). I had to remove some XClearArea calls from resize() . I had doubts about
them for a while: they were using outdated values. I also decided to have labels clear their
'cell' before redrawing and collapsed drawXmLlabel into drawLabel.
- ClearArea breaks relayout out of resize and removes all calls to
XClearArea from these functions, adding them to add/delete row/column.
This hopefully gets all that working. It also eliminates private header
dependence of some examples allowing to remove xbaeClearCell.
- CopyPerCell.patch replaces calls to xbaeCopyPerCell by xbaeCreatePerCell
where apropriate. After that there is a call to copypercell left in a dead
piece of code and another one that is commented out. So I killed the dead
code and the whloe xbaeCopyPerCell function.
- userWidget.patch: (at least) since the new XbaeMatrixSetUserWidget apeared,
user widgets can be treated just like the text widget and xbaeMoveUserWidget
is not needed anymore.
* Fixes from Michel Bardiaux :
- allow build using an older gcc (still the official compiler in Debian)
- fix a 'BadWindow' happening with the following sequence:
XbaeCreateMatrix
XtRealizeWidget(toplevel)
create a widget as child of matrix
XbaeMatrixSetCellWidget
* Another one from Tobias : The first part makes the scroll managers part of the clips,
and the second part isolates all the scrolling to the clips widgets.
* Change the signature of XbaeMatrixGetCellPixmap so it has a return code.
* Implement XbaeMatrixSortRows() as the same function as XbaeMatrixSort().
Maybe the latter should disappear.
* Document the new interfaces from Arcad.
* Include code for creating a DLL under the Exceed environment.
Until we figure out how to auto-detect Exceed, a flag should be passed to configure
to enable this.
Intermediate release 4.50.97
----------------------------
* A submission of code by Arcad Systemhaus for :
- Added XbaeMatrixSort(Widget, void *proc);
- Moved the cell string into the per_cell structure, to be able to sort.
- Fixes for multiline text display in cells.
- Added XbaeMatrixTrackCellCallback which fires everytime the cursor enters
a new cell (something entercell/leavecell do differently).
* Modified ("fixed" really) cell width calculation, as requested by Evgeny,
to mimic the behaviour described in the Motif documentation.
Test release 4.50.93
--------------------
* Lots of bugfixes and code cleanup by Tobias Oed, see ChangeLog.
Test release 4.50.91
--------------------
* Lots of bugfixes and code cleanup by Tobias Oed, see ChangeLog.
Release 4.50.5
--------------
* Fix for XmNautoFill in XbaeInput when the pattern begins with an
optional literal, such as [-]d[d][d][d][d] .
* Add the new resources introduced in the last months to the example
Builder Xcessory integration file in examples/builderXcessory.
* Bugfix for a clip window size problem that showed up when resizing the
window (e.g. resize examples/traversal/traversal to smaller than the
original and then bigger). Bug #702560.
* Some source code cleanup.
* Fixes by James Georgas for colour handling.
* Fixes by Van to eliminate slider size warnings (see bugs #823041 and #823037).
* Fixes by David Traill for both resize and scrollbar warnings.
* Add XmNXmColumnLabel resource to handle XmString as column labels.
* Fix a memory problem with row_heights.
4.50.3 was not an official release.
It was a test version distributed to parties interested in it.
Release 4.50.2
--------------
* A memory (double free) related bugfix.
* Two new resources (XmNhorzFill, XmNvertFill) were added to allow for
additional fill behaviour. This allows you to specify how the matrix
treats highlighting of the last row or column.
* The foreground and background resources are no longer being set on
cell widgets, as this appears to be strange behaviour.
* Bugfix related to incorrect refresh after deleting rows.
* Remove some C++ style comments.
* Fix some build problems.
* Move some stuff in the share/ directory at installation.
* Implement the XBAE_PRODUCTION symbol to compile Xbae with or without
debugging code.
* Improve resizing rows and columns.
* Avoid X Errors when setting clip geometry to 0.
* Implement showColumnArrows and showRowArrows resources.
* Fix incorrect behaviour when rapidly clicking with two different
mouse buttons.
* Fix nested comments and a missing prototype.
* Fix scrollbar warning problem.
* Changed the column width measurement.
Release 4.50
------------
This is a stable release, basically equivalent to 4.9.13.
Changes in 4.9.13
-----------------
* Fix the GROHTML build problem.
* Bugfix : when resizing first of trailing fixed rows, the VSB slider
was not updated and a blank area was visible between clip and bottomclip.
* Changed useXbaeInput default value to False.
Changes in 4.9.11
-----------------
* Sascha Göbel and Greg Shebert are now part of the Xbae development team.
* Many bugfixes to the scrolling (smooth scrolling now),
and to resizing row heights.
* Received code from Sasha Göbel.
Changes in 4.9.9
----------------
* Bug fixes related to redrawing.
* Enable dynamic row/column resizes again.
* Include images in release files again - this makes for a large package.
* Build HTML manual pages from their sources (the HTML is in doc/,
but the sources are in src/*.3.in).
Changes in 4.9.7
----------------
* Sascha Göbel contributed substantial improvement to scrolling.
* Bugfixes.
Changes in 4.9.5
----------------
* Improved internal debugging support (dmalloc, dbmalloc)
* Fix pattern check in XbaeInput
* Remove support for Motif <=1.1
* Bugfix by Phil Eccles to position cell widgets correctly when row/column
headers are active.
1.7.1:
Verbose mode is not started as daemon by default.
Check rc file before starting as daemon (so warning about
configuration files are visible).
1.7.0:
Add guile/scheme configurtion file style (Many thanks to Marcello
Mathias Herreshoff).
xdesktopwaves is a cellular automata setting the background of your X
Windows desktop under water. Windows and mouse are like ships on the
sea. Each movement of these ends up in moving water waves. You can
even have rain and/or storm stirring up the water.
make they use /sbin/shutdown under *BSD, instead of the poweroff and
reboot commands.
The patch is a hack (not to be submitted back to the upstream developers
without modification), but is the simplest thing we can do to avoid ugly
and hard to maintain patches. I'll submit a better, but more complex
one, to GNOME's Bugzilla when I have more time.
Bump PKGREVISION to 1.
