Hi,
Here is a patch to correct the handling of socket timeouts in the
NATIVE-LAYER branch. This patch relies on the select function (which is
not the most portable way). A few syscalls should also be protected that
way but they will come in a forthcoming patch (like connect).
ChangeLog entry:
On Tue, 09 May 2006 22:53:10 +0200
Robert Schuster [EMAIL PROTECTED] wrote:
Hi,
the following patch fixes the exception which Roman observed in PR24216.
Inspired by the cause for this exception I started to investigate how the
AbstractDocument class of the RI behaves in the similar methods
On Wednesday 10 May 2006 07:09, Tom Tromey wrote:
Raif == Raif S Naffah [EMAIL PROTECTED] writes:
...
Raif the longOnly print-help method is for commands which options do
not have Raif short-form syntax; e.g. jarsigner. those commands
signal their Raif intention with the Parser constructor
I implemented image rendering for the new Java2D. This was easier than I
thought. I implemented all draw*Image() methods, so that they finally
end up calling drawRenderedImageImpl() (this of course works only on
RenderableImage and RenderedImage so far), which then uses an ImagePaint
to fetch the
There was a small bug in AlphaPaintContext, which made composited
painting appear too transparent. The mistake was that I pre-multiplied
the alpha value to itself.
2006-05-10 Roman Kennke [EMAIL PROTECTED]
* gnu/java/awt/java2d/AlphaCompositeContext.java
(compose): Don't
It seems that the PaintContext implementations are expected to produce
rasters that start at (0,0) and not at (x,y). I fixed this in
ColorPaintContext. The GradientPaintContext already works like that (in
fact, David told me that this GradientPaintContext was tested to work on
Sun, and so I could
Ingo noticed some performance glitches when working with the default RGB
color model supplied by ColorModel.getRGBDefault(). This was caused by :
- creating a new instance every time
- DirectColorModel not optimized for sRGB
He wrote a small specialized SRGBColorModel, which directly accesses the
Ingo implemented a facility to adjust the priority of the AWT
EventDispatchThread. The priority can now be adjusted using the system
property gnu.awt.dispatchthread.priority, and defaults to the old value
NORM_PRIORITY+1.
2006-05-10 Roman Kennke [EMAIL PROTECTED]
Reported by Ingo
hello all,
this patch handles wrapping long header, footer and description texts
with the tools' getopt Parser and OptionGroup classes.
ok to commit?
cheers;
rsn
Index: OptionGroup.java
===
RCS file:
Hi all,
This commit removes some unnecessary ThreadGroup access checks
that were happening during Thread and ThreadGroup access checks.
This fixes the failures in the Mauve tests I committed earlier
today.
Cheers,
Gary
Index: ChangeLog
Here comes a couple of fixes and adjustments for the Swing peers.
2006-05-10 Roman Kennke [EMAIL PROTECTED]
* gnu/java/awt/peer/swing/SwingComponent.java:
Some API comment fixlets.
* gnu/java/awt/peer/swing/SwingComponentPeer.java:
(createImage): Create a
On Mon, 2006-05-01 at 16:39 -0400, Lillian Angel wrote:
After testing a timezone app, I noticed that SimpleTimeZone is a little
buggy. I fixed it up a bit... I imagine there are more bugs in this
code. It does work better now.
2006-05-01 Lillian Angel [EMAIL PROTECTED]
*
Hi Tom,
On Mon, 2006-05-01 at 16:32 -0600, Tom Tromey wrote:
It also fixes a systematic bug in the 'compact' methods
2006-05-01 Tom Tromey [EMAIL PROTECTED]
* java/nio/ByteBufferImpl.java (compact): Don't reset position
in empty case.
This does introduce a mauve regression.
Hi Audrius,
On Wed, 2006-04-26 at 14:28 +0200, Audrius Meskauskas wrote:
2006-04-26 Audrius Meskauskas [EMAIL PROTECTED]
* javax/swing/tree/DefaultTreeSelectionModel.java
(addSelectionPath): Event construction fix (old and new lead were
always the same).
Hi,
There was a small issue in the toCanonicalForm() method in
PlatformHelper, used to get the canonical form of a file : at least on
Unix, the call
new File(/).getCanonicalPath()
would return // instead of /.
So I added a little test at the beginning of this method to correct this.
I also
Sorry wrong patch attached.
Attached is the right one.
On Wed, 2006-05-10 at 13:12 -0400, Lillian Angel wrote:
I reverted this patch. It caused regressions.
I am going to look into the bug I had and try to think of a better
solution.
2006-05-10 Lillian Angel [EMAIL PROTECTED]
*
This patch (committed) fixes a minor problem I noticed in the MetalRadioButtonUI
class. The look up keys for the UIDefaults table shouldn't be hard-coded, they
should be constructed using the UI delegate's getPropertyPrefix():
2006-05-10 David Gilbert [EMAIL PROTECTED]
*
hello all,
the attached patch --already committed-- adds line-wrapping to long
header, footer and description texts in getopt Parser and OptionGroup
classes.
2006-05-11 Raif S. Naffah [EMAIL PROTECTED]
* tools/gnu/classpath/tools/getopt/OptionGroup.java
(FILLER): New
It seems my last patch for JRootPane was wrong. I revert it, though it
causes http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27481 to be reopened.
2006-05-10 Roman Kennke [EMAIL PROTECTED]
* javax/swing/JRootPane.java
(createContentPane): Don't set background to null.
/Roman
Thanks to Wolfgang's work, we can eliminate this particular stub:
2006-05-10 Sven de Marothy [EMAIL PROTECTED]
* java/awt/print/PrinterJob.java:
(lookupPrintServices): Un-comment-out.
Index: java/awt/print/PrinterJob.java
This sets the bg color of the content pane of a JInternalFrame to null
at UI install time, and only if there is no custom color already set.
This fixes bug#27481 and causes no regressions AFAICS.
2006-05-10 Roman Kennke [EMAIL PROTECTED]
PR 27481
*
This patch (committed) marks some stub methods in this class:
2006-05-10 David Gilbert [EMAIL PROTECTED]
* javax/swing/TransferHandler.java: Marked stub methods.
Regards,
Dave
Index: javax/swing/TransferHandler.java
===
22 matches
Mail list logo