redefines about which buildlink.mk files would care is BUILDLINK_X11_DIR,
which points to the location of the X11R6 hierarchy used during building.
If x11.buildlink.mk isn't included, then BUILDLINK_X11_DIR defaults to
${X11BASE} (set in bsd.pkg.mk), so its value is always safe to use. Remove
the ifdefs surrounding the use of BUILDLINK_X11_DIR in tk/buildlink.mk and
revert changes to move x11.buildlink.mk before the other buildlink.mk files.
use X11_BUILDLINK_MK as a test value. Generally just reordering the
inclusions so that x11.buildlink.mk comes before the other buildlink.mk
files will make everthing work.
to ${X11BASE} in the header and library search paths into references to
${LOCALBASE}/share/x11-links. These packages should now be strongly-
buildlinked regardless of whether xpkgwedge is installed.
Changes well-tested on NetBSD-1.5X/i386 with and without xpkgwedge and
lightly-tested on NetBSD-1.5.1/alpha without xpkgwedge.
BUILDLINK_PREFIX.<pkgname>. This allows buildlink to find X11BASE packages
regardless of whether they were installed before or after xpkgwedge was
installed. Idea by Alistair Crooks <agc@pkgsrc.org>.
FOO_REQD=1.0 being converted to foo>=1.0, one can now directly specify
the dependency pattern as FOO_DEPENDS=foo>=1.0. This allows things like
JPEG_DEPENDS=jpeg-6b, or fancier expressions like for postgresql-lib.
Change existing FOO_REQD definitions in Makefiles to FOO_DEPENDS.
CXXFLAGS, and LDFLAGS by the buildlink.mk files so remove the extra
definitions to add them from the package Makefiles. As advised by the
bsd.buildlink.mk file, also ensure that the buildlink.mk files are
included prior to defining any package-specific CFLAGS/LDFLAGS to ensure
that the buildlink directories are at the head of the compiler search
paths.
linked from a particular package, and add a pre-configure target to
the buildlink.mk file to more painlessly use buildlink.mk files. A
${BUILDLINK_TARGETS} variable still exists in case a package _must_
define NO_CONFIGURE.
- Added new const const pointer versions of pixmap functions to eliminate
an annoying pointer warning message that was generated by the Sun and
other C++ compilers.
- Eliminated all "var hides class::var" warnings.
- Eliminated all "string literal converted to char *" warnings.
- The file chooser buttons use user_data() rather than the label to decide
what to do, allowing the label to be somewhat cleaner.
- Fl_has_idle only tested N-1 callbacks and missed one.
- Fl_Gl_Window has a new method to allow you to get and set the context:
void Fl_Gl_Window::context(void*, int destroy = 0)
void* Fl_Gl_Window::context() const;
Return or set a pointer to the GLContext that this window is using. This
is a system-dependent structure, but it is portable to copy the context
from one window to another. You can also set it to NULL, which will
force FLTK to recreate the context the next time make_current() is
called, this is useful for getting around bugs in OpenGL implementations.
If destroy_flag is true the context will be destroyed by fltk when the
window is destroyed, or when the mode() is changed, or the next time
context(x) is called.
- Added Fl::add_check(...), Fl::remove_check, and Fl::has_check. These are
similar to idle callbacks but are only called just before it waits for
new events. They can be used to watch for changes in global state and
respond to them.
- Added simple FLTK and FLUID manual pages.
out of date - it was based on a.out OBJECT_FMT, and added entries in the
generated PLISTs to reflect the symlinks that ELF packages uses. It also
tried to be clever, and removed and recreated any symbolic links that were
created, which has resulted in some fun, especially with packages which
use dlopen(3) to load modules. Some recent changes to our ld.so to bring
it more into line with other Operating Systems also exposed some cracks.
+ Modify bsd.pkg.mk and its shared object handling, so that PLISTs now contain
the ELF symlinks.
+ Don't mess about with file system entries when handling shared objects in
bsd.pkg.mk, since it's likely that libtool and the BSD *.mk processing will
have got it right, and have a much better idea than we do.
+ Modify PLISTs to contain "ELF symlinks"
+ On a.out platforms, delete any "ELF symlinks" from the generated PLISTs
+ On ELF platforms, no extra processing needs to be done in bsd.pkg.mk
+ Modify print-PLIST target in bsd.pkg.mk to add dummy symlink entries on
a.out platforms
+ Update the documentation in Packages.txt
With many thanks to Thomas Klausner for keeping me honest with this.
during a normal build to reduce build times on slower platforms. Relevant
changes from version 1.0.9 include various bug fixes, less memory consumption
for colors, improvements to the code generation in fluid, and improvements to
the way scrollbars and buttons are handled. There is also the following GL
optimization for certain OpenGL implementations:
- Performance of Fl_Gl_Window may be improved on some
types of OpenGL implementations, in particular MESA
or other software emulators, by setting the
GL_SWAP_TYPE environment variable. This variable
declares what is in the back buffer after you do a
swapbuffers:
setenv GL_SWAP_TYPE COPY
This indicates that the back buffer is copied to
the front buffer, and still contains it's old
data. This is true of many hardware
implementations. Setting this will speed up
emulation of overlays, and widgets that can do
partial update can take advantage of this as
damage() will not be cleared to -1.
setenv GL_SWAP_TYPE NODAMAGE
This indicates that nothing changes the back
buffer except drawing into it. This is true of
MESA and Win32 software emulation and perhaps some
hardware emulation on systems with lots of memory.
All other values for GL_SWAP_TYPE, and not setting
the variable, cause fltk to assumme that the back
buffer must be completely redrawn after a swap.
This is easily tested by running the gl_overlay demo
program and seeing if the display is correct when
you drag another window over it or if you drag the
window off the screen and back on. You have to exit
and run the program again for it to see any changes
to the environment variable.
Add a new USE_LIBTOOL definition that uses the libtool package instead of
pkglibtool which is now considered outdated.
USE_PKGLIBTOOL is available for backwards compatibility with old packages
but is deprecated for new packages.
CHANGES SINCE FLTK 1.0.6
- Fixed Fl_Input_ bug under WIN32 - no longer stop accepting input
when one of the "Windows" keys is pressed.
- Now call TranslateEvent for all events under WIN32.
- Fixes for OpenBSD and NetBSD
- The FL_CURSOR_HAND cursor now uses the IDC_HAND cursor instead of
IDC_UPARROW under Windows 98 and 2000.
- Fl_Scrollbar now does a page-up/down when you click outside the
scroller.
- Fl_Window::show(0, NULL) causes core dump
- Fixed a compile-time error in fl_call_main.c for Borland C++.
- "fluid -c filename.fl" would try to open an X display if the
FLUID file contained an Fl_Browser widget.
- Fl_Browser now correctly measures items with @C or @B color
formatting commands.
- Fixed a bitmap drawing bug for WIN32 (bit reversal table was wrong)
- fl_xyz() dialogs now set a title in the title bar.
- fl_alert() sounds the bell under X11.
- fl_xyz() dialogs now call MessageBeep() under WIN32.
- Fl_Browser_ didn't draw the selection box with the inactive color
when the browser wasn't activated.
- Fl_Browser now responds to FL_KEYBOARD as well as FL_SHORTCUT. If
you subclass it to accept focus then keyboard navigation will work.
- Fl_Tile and Fl_Tabs do their callback when the user changes their
display.
- Made some of the private methods of Fl_Browser protected.
- Now set win_gravity correctly, this helps some X
window managers that use it position the window where
FLTK wants it to be.
- 0-width browsers crashed.
- Minor change: if the X window manager does not do
anything else with windows that don't have their
position specified, the windows appear centered in the
screen, rather than in the top-left corner. This
happened with modal windows under Irix 4Dwm. This
also causes windows to be centered when no window
manager is running, which might be useful for
installation gui programs?
- Clicking in an Fl_Input field the first time selects the entire
field.
- Clicking the middle mouse button in an Fl_Input field now inserts
the text at the indicated position instead of the cursor position.
- Drag-selecting text in an Fl_Input field now copies the text
automatically.
- Fl::flush() no longer calls the draw() method for invisible windows.
- Calling deactivate() on an invisible widget could cause an
infinite loop in some obscure cases.
- Added #pragma's for SGI C++ compilers - the 6.{23} X headers had
errors in them.
- Fl_Gl_Window::ortho() changed so that text and images
are not erased if the origin is off the left/bottom of the
window.
- Small change to Fl_Input so that a click that gives it
the focus also selects all the text.
- Fixed a slider drawing problem.
- You can now add/delete children of Fl_Tabs widgets whether or
not they are visible.
- Now embed woff options for SGI C++ compilers (gets rid of X11
header warnings)
- draw_pixmap used a cast that the Digital UNIX C++ compiler didn't
like.
- The GLUT function key constants were off by one.
- The XPM reading code didn't handle RGB colors other than #rrggbb.
CHANGES SINCE FLTK 1.0.5
- Fl_win32.cxx defined WM_MOUSE_LEAVE instead of WM_MOUSELEAVE.
- Fl_get_key_win32.cxx needed to include <ctype.h>
- gl_draw_pixmap.cxx needed a pointer cast for ANSI C++.
- Fl_Repeat_Button didn't always delete its timeout.
- Now keep track of the current OpenGL context; this provides
significant performance improvements for OpenGL applications
with a single context.
CHANGES SINCE FLTK 1.0.4
- Fl_Roller didn't handle a width and height of 0.
- filename_list() fix for FreeBSD.
- Fixed RPM install docos - needed "--install" option...
- Fl_Browser_ wouldn't draw the vertical scrollbar right away if it
added a horizontal one which covered the last line.
- Fl_Tabs problems - single-character labels don't show up (problem in
measure_tabs() or measure_label() methods?), and doesn't clear top
tab area before drawing tabs.
- Fl_Browser needs a destructor.
- fl_draw_label() quoted characters between 0x80 and 0xa0, which
caused problems for some programs using the WinANSI character set.
- FLUID didn't handle declared class destructors.
- Fixed another WIN32 cut/paste bug.
- Fl_Tabs didn't work properly when there was only 1 tab.
- Fl_Menu::add() didn't delete the old array.
- Fl_Repeat_Button didn't delete its timeout when disabled.
- fl_draw() would crash if no font was set (now defaults to
a 14-pixel Helvetica font)
- Can't forward declare classes; need to check for "class ", "struct ",
"union ", etc. See Bill's message
- Added #pragma around xlib.h for IRIX
- FL_KEYBOARD events have the correct x/y when sent to child X
windows. Note that if you worked around this bug by adjusting the
x/y yourself you will have to change your code. In addition all
events have the correct x/y when sent to the grab() widget. And
the code to do all this was simplified a lot.
- The XPM code didn't handle named colors with spaces in the names.
- Pressing ESCape closed the window with pointer focus, even if there
was a modal window open (now closes the modal window).
- Fluid no longer produces trigraphs accidentally in the image data.
- Fluid uses string constant concatenation to produce shorter image
data.
- The Fl_Group deletion code crashed if there was exactly one child
widget.
- Simulated overlays in single-buffered Fl_Gl_Windows now draw
correctly (though very slowly as it requires the entire window to
be redrawn to erase the overlay). This fix ported our Digital
Domain programs better to systems with no overlay hardware.
- Added support for extern "C" declarations in FLUID.
- Added Fl_Pack support to FLUID.
- Fixed the order of #include's in FLUID generated header files.
- Fixed detection of vsnprintf and snprintf under HP-UX 10.20 once
and for all.
- The checkers demo did not compile with GCC 2.95
- FLUID didn't output virtual destructors properly.
- Added inline "make_visible()" method to Fl_Browser.
- Fl::wait() now returns immediately if any timeouts are
called.
- 16-bit XPM files are now properly handled.
- Fl_Window::resize() was missing FL_EXPORT (caused problems
with Windows DLLs)
- FLUID was writing extern declarations twice.
- New FLUID arrow key functionality: arrows move by one pixel, shift+arrow
resizes, ctrl+arrow steps by grid
CHANGES SINCE FLTK 1.0.3
- Documentation updates
- Fl_Browser::bottomline(size) didn't scroll to the bottom
if the second-to-last line was visible.
- fl_wait() didn't poll FDs properly for WIN32.
- Fixed DLL definitions for BC++.
- FLUID now handles nested classes properly.
- The "connect" demo now does a wait() for the PPP process
so that you aren't left with a lot of zombie processes.
- Fixed the FLTK colormap to use FF instead of F4 for full
intensity values.
- Minor change to scrollbar drawing code to match other
toolkits.
- New selections would cancel themselves out in WIN32.
- The header file links were broken in the IRIX
distributions.
- fl_elapsed() now always uses GetClockTick() for WIN32.
- fl_display is now initialized to GetModuleHandle(NULL) -
this fixes problems people had with Cygwin and MingW32.
- WinMain() is no longer compiled in with Cygwin and
MingW32; it wasn't being used for those compilers anyways.
- Added Solaris compiler options to configure script.
- Fl_Value_Input wouldn't update properly if you set the
value from a callback.
- Fl_Tile wouldn't resize if the resizeable widget was the
last child.
- Was missing #include <ctype.h> and #include <stdlib.h> in
several files, which caused problems on some platforms.
- Fixed another case where Fl_Browser_ could get in an
infinite resizing loop.
- Fl_win32.cxx now includes <FL/filename.H> to export missing
DLL symbols.
- Fluid didn't handle member functions that include the
scope operator.
- Fl_Chart was dividing by 0 if there were no data samples
or if they were all the same (min == max).
