Revert a portion of r32955 which I committed last april but I now
see was a mistake, and I just discovered is breaking the copy-on-scroll
behaviour of NSClipView - we end up always redrawing the entire visible
portion of the document view right now.
In r32955 I added [self setNeedsDisplay: YES] calls to these methods
even though these are documented explicitly as not marking the view for
needing display. Approved by Fred for commit during the code freeze.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@34606 72102866-910b-0410-8b05-ffd578937521
by the placement value. Previously, we would translate only if the
rect parameter was empty, which doesn't make sense to me.
This fixes the bug where the last page of a document printed from
Ink/TextEdit was incorrectly positioned at the bottom of the
printed page.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@34352 72102866-910b-0410-8b05-ffd578937521
and unsigned with NSUInteger. This allows to recompile gui on
64 bit systems after the change to NSNotFound in base.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@34312 72102866-910b-0410-8b05-ffd578937521
Note that this makes the NSView autoresizing and rounding test pass,
but the new centerScanRect test I just added fail.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@34104 72102866-910b-0410-8b05-ffd578937521
(if necessary) _invalidRect so it lies on integral device pixels.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33809 72102866-910b-0410-8b05-ffd578937521
autoresizing logic into a function. Add an extra code path to handle
when the flexible space is 0, which behaves like the old version
of the code (it divides the available space evenly).
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33490 72102866-910b-0410-8b05-ffd578937521
* Headers/AppKit/NSAffineTransform.h,
* Headers/AppKit/NSView.h,
* Source/NSAffineTransform.m,
* Source/NSPrintOperation.m,
* Source/NSView.m: Use CGFlaot instead of float.
* Tests/gui/NSView/NSView_bounds_scale.m: New test file that currently fails.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33398 72102866-910b-0410-8b05-ffd578937521
the coordinate transformations and the cursor rects before the
window test. Remove calls to coordinate invalidation from other places.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33347 72102866-910b-0410-8b05-ffd578937521
methods plus a few helpers. We pass now all the frame/bounds test cases.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33343 72102866-910b-0410-8b05-ffd578937521
* Source/NSView.m (-convertPoint:toView:): Correct this
method; it was applying the transforms backwards :-(
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@33001 72102866-910b-0410-8b05-ffd578937521
Fix a bug where updateBoundsMatrix would refuse to restore a
view's scale to its original value, after a modified bounds rect
had been set.
* Source/NSView.m (-setBounds:, -setBoundsOrigin:, -setBoundsSize:):
Call [self setNeedsDisplay: YES]; at the ends of these methods.
While the Cocoa docs explicitly say that calling these methods
won't mark the view as needing redisplay, this doesn't seem to
be true in practice.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@32955 72102866-910b-0410-8b05-ffd578937521
the window content view's superview's transformation matrix is
the identity matrix.
* Source/NSView.m (-convertPoint..., convertRect..., convertSize...):
Rewrite these methods to remove the same assumption as above, and
also call -_matrixToWindow/-_matrixFromWindow rather than accessing
ivars directly.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@32887 72102866-910b-0410-8b05-ffd578937521
retained in the binding table.
Leak found by static code analyser.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@32484 72102866-910b-0410-8b05-ffd578937521
[-releaseGState] make sure to keep the value of _allocate_gstate.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@31409 72102866-910b-0410-8b05-ffd578937521
Patch by Georg Fleischmann <G.Fleischmann@vhf.de>.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@31350 72102866-910b-0410-8b05-ffd578937521
_rFlagsType.flipped_view.
Still needs some work to automatically invalidate coordinates when the
flipping changes e.g. between two drawing sessions.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@31233 72102866-910b-0410-8b05-ffd578937521
See bug report #27782
In particular, fixed -[NSImage drawXXX] and -[NSImage composite/dissolveXXX]
methods to work exactly as Cocoa when the Cairo backend is used.
Added a new draw operator (in addition to composite) to the backend. Cairo is
the only backend that implements it for now.
Eliminated as many flipping checks as possible.
Warning: Untested with the winlib backend.
You must update, recompile and install both Back and Gui.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@30523 72102866-910b-0410-8b05-ffd578937521
plenty of warnings and the result isn't working at least on 64 bit
machines :-(
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@29685 72102866-910b-0410-8b05-ffd578937521
call -setNeedsDisplay:YES on the superview, rather than self
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@29417 72102866-910b-0410-8b05-ffd578937521
from nib's, with subviews. The subview coordinates just need to be
invalidated when the view is unhidden.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@29416 72102866-910b-0410-8b05-ffd578937521
_post_frame_changes and _post_bounds_changes to YES by default.
Patch by Quentin Mathe <qmathe@gmail.com>.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@28465 72102866-910b-0410-8b05-ffd578937521
views as needing display, even when the invalidated rectangle
didn't change.
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/gui/trunk@28382 72102866-910b-0410-8b05-ffd578937521
