]> source.dussan.org Git - tigervnc.git/log
tigervnc.git
10 years agoDocument test cases for testing the keyboard handling
Pierre Ossman [Wed, 17 Sep 2014 11:55:51 +0000 (13:55 +0200)]
Document test cases for testing the keyboard handling

10 years agoNone of these files are relevant today
Pierre Ossman [Wed, 17 Sep 2014 11:55:18 +0000 (13:55 +0200)]
None of these files are relevant today

10 years agoMoved release documentation to the wiki
Pierre Ossman [Wed, 17 Sep 2014 11:54:28 +0000 (13:54 +0200)]
Moved release documentation to the wiki

10 years agoFix broken FLTK patch 31/head
Pierre Ossman [Tue, 16 Sep 2014 12:52:26 +0000 (14:52 +0200)]
Fix broken FLTK patch

Remove some lines that weren't consistent with the earlier
suggested patches.

10 years agoMerge branch 'xhandlers' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 15 Sep 2014 11:43:16 +0000 (13:43 +0200)]
Merge branch 'xhandlers' of https://github.com/CendioOssman/tigervnc

10 years agoUpdate FLTK patches based on feeback from upstream 30/head
Pierre Ossman [Thu, 11 Sep 2014 08:48:29 +0000 (10:48 +0200)]
Update FLTK patches based on feeback from upstream

10 years agoProtect the optional IM disabling calls with #ifdef:s
Pierre Ossman [Thu, 11 Sep 2014 08:46:12 +0000 (10:46 +0200)]
Protect the optional IM disabling calls with #ifdef:s

10 years agoAdded missing part of write block handler. 29/head
Tim Waugh [Tue, 9 Sep 2014 13:40:17 +0000 (14:40 +0100)]
Added missing part of write block handler.

https://bugzilla.redhat.com/show_bug.cgi?id=1137023

10 years agoMerge pull request #28 from bphinz/buildfixes
bphinz [Sun, 7 Sep 2014 14:14:33 +0000 (10:14 -0400)]
Merge pull request #28 from bphinz/buildfixes

Removed patch that was merged into trunk in 6bc1de83a7

10 years agoRemoved patch that was merged into trunk in 6bc1de83a7 28/head
Brian P. Hinz [Sun, 7 Sep 2014 14:05:09 +0000 (10:05 -0400)]
Removed patch that was merged into trunk in 6bc1de83a7

10 years agoMerge branch 'xorgserver116' of https://github.com/twaugh/tigervnc
Pierre Ossman [Fri, 5 Sep 2014 08:22:38 +0000 (10:22 +0200)]
Merge branch 'xorgserver116' of https://github.com/twaugh/tigervnc

10 years agoFix linking when dri3 and present are enabled. 26/head
Tim Waugh [Thu, 4 Sep 2014 15:25:12 +0000 (16:25 +0100)]
Fix linking when dri3 and present are enabled.

10 years agoNo need to declare 'display' here either.
Tim Waugh [Thu, 4 Sep 2014 13:35:05 +0000 (14:35 +0100)]
No need to declare 'display' here either.

10 years agoGCFuncs and GCOps are const since 1.16.
Tim Waugh [Thu, 4 Sep 2014 11:36:09 +0000 (12:36 +0100)]
GCFuncs and GCOps are const since 1.16.

10 years agoNo need to declare 'display' here.
Tim Waugh [Thu, 4 Sep 2014 10:41:01 +0000 (11:41 +0100)]
No need to declare 'display' here.

10 years agoAnother LoadExtensionList() that needs an #ifdef.
Tim Waugh [Thu, 4 Sep 2014 09:52:06 +0000 (10:52 +0100)]
Another LoadExtensionList() that needs an #ifdef.

10 years agoLoadExtensionList is new in 1.16.
Tim Waugh [Wed, 3 Sep 2014 14:48:14 +0000 (15:48 +0100)]
LoadExtensionList is new in 1.16.

10 years agoUpdated to build against xorg-x11-server 1.16.0.
Tim Waugh [Wed, 3 Sep 2014 11:38:15 +0000 (12:38 +0100)]
Updated to build against xorg-x11-server 1.16.0.

10 years agoUse different keysyms for left and right alt on OS X
Pierre Ossman [Thu, 21 Aug 2014 11:44:28 +0000 (13:44 +0200)]
Use different keysyms for left and right alt on OS X

The server will get them confused otherwise if they are pressed
at the same time.

10 years agoDisable input methods when the viewport is focused
Pierre Ossman [Tue, 19 Aug 2014 12:08:04 +0000 (14:08 +0200)]
Disable input methods when the viewport is focused

Input methods are way too complex for us to map them to the VNC
protocol in any sane manner. Best just to disable them and rely
on simple keyboard behaviour when the viewport is active.

10 years agoIgnore FLTK events when we have low level hooks
Pierre Ossman [Tue, 19 Aug 2014 12:06:53 +0000 (14:06 +0200)]
Ignore FLTK events when we have low level hooks

Some events can sneak through, e.g. from some virtual keyboards.
We need to ignore these to avoid confusion.

10 years agoAdd OS X keyboard handler
Pierre Ossman [Mon, 21 Jul 2014 14:48:43 +0000 (16:48 +0200)]
Add OS X keyboard handler

10 years agoAdd Windows keyboard handler
Pierre Ossman [Mon, 21 Jul 2014 14:46:22 +0000 (16:46 +0200)]
Add Windows keyboard handler

10 years agoAdd X11 keyboard handler
Pierre Ossman [Mon, 21 Jul 2014 14:42:12 +0000 (16:42 +0200)]
Add X11 keyboard handler

10 years agoAdd xhandler hook
Pierre Ossman [Fri, 22 Aug 2014 13:10:22 +0000 (15:10 +0200)]
Add xhandler hook

Boiler plate code to intercept system events from FLTK so that
we can generate proper keyboard messages.

10 years agoAdd helper to get combining codepoint from a spacing character
Pierre Ossman [Mon, 21 Jul 2014 14:36:50 +0000 (16:36 +0200)]
Add helper to get combining codepoint from a spacing character

10 years agoBetter codepoint for combining ypogegrammeni (dead iota)
Pierre Ossman [Mon, 21 Jul 2014 14:35:51 +0000 (16:35 +0200)]
Better codepoint for combining ypogegrammeni (dead iota)

10 years agoAttempt to handle Ctrl+key
Pierre Ossman [Fri, 22 Aug 2014 13:09:54 +0000 (15:09 +0200)]
Attempt to handle Ctrl+key

Hacky attempt to try to figure out the actual key pressed when
FLTK is giving us control codes.

10 years agoRemove simple keyboard model
Pierre Ossman [Mon, 21 Jul 2014 14:30:08 +0000 (16:30 +0200)]
Remove simple keyboard model

Remove the FLTK simple keyboard system and reorganise things in
preparation for a more direct approach.

10 years agoShip FLTK patches instead of referring a URI
Pierre Ossman [Fri, 22 Aug 2014 12:43:33 +0000 (14:43 +0200)]
Ship FLTK patches instead of referring a URI

It makes things more self contained (although you still need to
download FLTK itself), avoids the confusion of STR:s in varying
states, and allows us to refer to patches that aren't available
on an STR.

10 years agoUpdate FLTK requirement
Pierre Ossman [Fri, 22 Aug 2014 12:41:03 +0000 (14:41 +0200)]
Update FLTK requirement

We actually require FLTK 1.3.2 as we reference FL_SOCKET.

10 years agoMerge pull request #22 from bphinz/i18nfix
bphinz [Thu, 21 Aug 2014 00:48:28 +0000 (20:48 -0400)]
Merge pull request #22 from bphinz/i18nfix

Fix some cmake quoting and escape issues.  The un-escaped parens generat...

10 years agoFix some cmake quoting and escape issues. The un-escaped parens generate a "Argument... 22/head
Brian P. Hinz [Fri, 8 Aug 2014 03:52:19 +0000 (23:52 -0400)]
Fix some cmake quoting and escape issues.  The un-escaped parens generate a "Argument not separated from preceding token by whitespace" warning. The '--directory' argument might need to be quoted if PROJECT_SOURCE_DIR contains spaces.

10 years agoMerge pull request #21 from bphinz/precise-damage-fix
bphinz [Fri, 8 Aug 2014 03:31:35 +0000 (23:31 -0400)]
Merge pull request #21 from bphinz/precise-damage-fix

Added missing dependency in Ubuntu precise package that caused x0vncserver to fall back to polling if the build host didn't already have libxdamage-dev installed.

10 years ago\nAdded missing dependency that caused x0vncserver to fall back to polling 21/head
Brian P. Hinz [Fri, 8 Aug 2014 03:27:52 +0000 (23:27 -0400)]
\nAdded missing dependency that caused x0vncserver to fall back to polling

10 years agoWindows debug builds of vncviewer should have a console
Pierre Ossman [Mon, 21 Jul 2014 14:22:43 +0000 (16:22 +0200)]
Windows debug builds of vncviewer should have a console

Makes it easier to debug it by having convenient access to the
log output.

10 years agoUse NoSymbol as the error code
Pierre Ossman [Mon, 21 Jul 2014 14:17:23 +0000 (16:17 +0200)]
Use NoSymbol as the error code

This is how the X11 API does things, so it's clearer if we use
the same principle.

10 years agoMake keysymtab static and const
Pierre Ossman [Mon, 21 Jul 2014 14:12:00 +0000 (16:12 +0200)]
Make keysymtab static and const

Static to avoid namespace polution, and const so that it can be
shared among instances.

10 years agoUse printf constructor safely
Pierre Ossman [Mon, 21 Jul 2014 14:10:03 +0000 (16:10 +0200)]
Use printf constructor safely

The input string might contain formatting marks, so make sure those
are ignored.

10 years agoMerge branch 'pointersync' of https://github.com/twaugh/tigervnc
Pierre Ossman [Tue, 15 Jul 2014 07:33:34 +0000 (09:33 +0200)]
Merge branch 'pointersync' of https://github.com/twaugh/tigervnc

10 years agoMerge pull request #14 from bphinz/master
bphinz [Mon, 14 Jul 2014 22:09:35 +0000 (18:09 -0400)]
Merge pull request #14 from bphinz/master

Remove OpenSuse clean-pressed-key patch

10 years agoRemove OpenSuse clean-pressed-key patch 14/head 17/head
Brian P. Hinz [Mon, 14 Jul 2014 21:56:23 +0000 (17:56 -0400)]
Remove OpenSuse clean-pressed-key patch

Remove clean-pressed-key-on-exit inherited from original spec file that
ours was derived from.

10 years agoMerge pull request #13 from bphinz/master
bphinz [Mon, 14 Jul 2014 21:24:57 +0000 (17:24 -0400)]
Merge pull request #13 from bphinz/master

Remove OpenSuse newfbsize patch

10 years agoRemove OpenSuse newfbsize patch 13/head
Brian P. Hinz [Mon, 14 Jul 2014 21:15:12 +0000 (17:15 -0400)]
Remove OpenSuse newfbsize patch

Remove patch inherited from Suse RPM that our spec was derived from.

10 years agoUpdate the cursor position using GetSpritePosition(). 12/head
Tim Waugh [Mon, 14 Jul 2014 16:04:56 +0000 (17:04 +0100)]
Update the cursor position using GetSpritePosition().

The module needs to correctly track the mouse position by querying the
current X cursor position in InputDevice::getPointerPos().

Fixes: https://sourceforge.net/p/tigervnc/bug-tracker/152/
10 years agoMerge branch 'encodemanager' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 14 Jul 2014 14:22:05 +0000 (16:22 +0200)]
Merge branch 'encodemanager' of https://github.com/CendioOssman/tigervnc

10 years agoMerge branch 'nopixeltrans' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 14 Jul 2014 14:17:46 +0000 (16:17 +0200)]
Merge branch 'nopixeltrans' of https://github.com/CendioOssman/tigervnc

10 years agoMerge branch 'nocolourmap' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 14 Jul 2014 14:15:16 +0000 (16:15 +0200)]
Merge branch 'nocolourmap' of https://github.com/CendioOssman/tigervnc

10 years agoString literals are constant 11/head
Pierre Ossman [Wed, 9 Jul 2014 15:02:27 +0000 (17:02 +0200)]
String literals are constant

10 years agoRemove PixelTransformer and TransImageGetter 10/head
Pierre Ossman [Wed, 9 Jul 2014 14:49:16 +0000 (16:49 +0200)]
Remove PixelTransformer and TransImageGetter

There is nothing using these classes anymore, so remove the clutter.

10 years agoMove image encoding logic into a central EncodeManager class
Pierre Ossman [Fri, 14 Mar 2014 14:59:46 +0000 (15:59 +0100)]
Move image encoding logic into a central EncodeManager class

This allows us to apply a lot more server logic
independently of which encoder is in use.

Most of this class are things moved over from the
Tight encoder.

10 years agoUse PixelBuffer objects as the interface for encoders and decoders
Pierre Ossman [Wed, 9 Jul 2014 14:44:11 +0000 (16:44 +0200)]
Use PixelBuffer objects as the interface for encoders and decoders

This avoid a lot of unnecessary middle men. This also pushes the
responsibility for pixel format conversion into the encoders and
decoders. The new bufferFromBuffer() is used for direct conversion,
rather than PixelTransformer/TransImageGetter.

10 years agoStore the mouse cursor in the ConnParams object
Pierre Ossman [Thu, 13 Feb 2014 13:40:25 +0000 (14:40 +0100)]
Store the mouse cursor in the ConnParams object

Like we do for everything else. This also gets rid
of the callback, which is a bit out of place
compared to everything else.

10 years agoAllow PixelBuffers to be const
Pierre Ossman [Thu, 13 Feb 2014 13:37:25 +0000 (14:37 +0100)]
Allow PixelBuffers to be const

10 years agoAdd helper class for a rendered cursor
Pierre Ossman [Wed, 12 Feb 2014 15:33:43 +0000 (16:33 +0100)]
Add helper class for a rendered cursor

Add a magical cursor framebuffer class for handling when you
want to render the cursor on the server side. Keeps the cursor
specific magic in one contained place.

10 years agoAdd convenience variant of getImage()
Pierre Ossman [Wed, 12 Feb 2014 12:24:15 +0000 (13:24 +0100)]
Add convenience variant of getImage()

10 years agoMerge branch 'inputreset' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 14 Jul 2014 13:03:05 +0000 (15:03 +0200)]
Merge branch 'inputreset' of https://github.com/CendioOssman/tigervnc

10 years agoHandle closing of input devices 6/head
Pierre Ossman [Mon, 14 Jul 2014 11:46:48 +0000 (13:46 +0200)]
Handle closing of input devices

The core can decide to close an input device (e.g. during a
server reset). Make sure we respect this by forgetting about
the device and recreating it again later.

10 years agoRemove private storage from InputDevice
Pierre Ossman [Mon, 14 Jul 2014 11:44:46 +0000 (13:44 +0200)]
Remove private storage from InputDevice

We don't need this anymore as we now only have a single
InputDevice object and can easily refer to it directly.

10 years agoMake InputDevice into a singleton
Pierre Ossman [Mon, 14 Jul 2014 11:41:36 +0000 (13:41 +0200)]
Make InputDevice into a singleton

Xorg input devices are global so it doesn't make sense that we
have one InputDevice object per XserverDesktop.

10 years agoMove all functions that InputDevice uses into the class
Pierre Ossman [Mon, 14 Jul 2014 11:39:57 +0000 (13:39 +0200)]
Move all functions that InputDevice uses into the class

10 years agoLet XserverDesktop handle syncing cursor position
Pierre Ossman [Mon, 14 Jul 2014 11:36:47 +0000 (13:36 +0200)]
Let XserverDesktop handle syncing cursor position

This makes InputDevice cleaner and just a glue layer for Xorg
input devices.

10 years agoRe-init VNC extension on server reset
Pierre Ossman [Mon, 14 Jul 2014 11:20:24 +0000 (13:20 +0200)]
Re-init VNC extension on server reset

fc23895 blocked out a bit too much code when trying to make
libvnc handle server resets. Even though configuration should
only be handled once, the underlying VNC extension needs to
be reinitilised as the X core has removed all extensions on
reset.

10 years agoWe need to set a label on all pointer buttons
Pierre Ossman [Fri, 11 Jul 2014 12:03:44 +0000 (14:03 +0200)]
We need to set a label on all pointer buttons

10 years agoAdd convenience functions to ModifiablePixelBuffer
Pierre Ossman [Thu, 13 Feb 2014 09:38:48 +0000 (10:38 +0100)]
Add convenience functions to ModifiablePixelBuffer

Allows you to modify the buffer with data in a
different pixel format.

10 years agoTest results comparing PixelTransformer and bufferFromBuffer()
Pierre Ossman [Wed, 9 Jul 2014 14:28:55 +0000 (16:28 +0200)]
Test results comparing PixelTransformer and bufferFromBuffer()

10 years agoAdd optimised buffer conversion
Pierre Ossman [Thu, 30 Jan 2014 09:47:07 +0000 (10:47 +0100)]
Add optimised buffer conversion

Handles the common cases when the target or source are in
the preferred 888 format.

If one of the buffers is not 888, then it must also be properly
aligned (which is commonly the case). Performance is now in many
cases on par with PixelTransformer.

10 years agoAdd ability to directly convert between two pixel formats
Pierre Ossman [Wed, 29 Jan 2014 16:00:36 +0000 (17:00 +0100)]
Add ability to directly convert between two pixel formats

This is a lot easier and cheaper than having to set up a complete
PixelTransformer object.

10 years agoUse lookup tables for pixel upconversion
Pierre Ossman [Tue, 8 Jul 2014 13:06:25 +0000 (15:06 +0200)]
Use lookup tables for pixel upconversion

You have to loop many times when converting e.g. 2 bits to 8,
so have a couple of lookup tables to speed things up. We still
use bit replication to get from 8 bits to 16 bits though.

10 years agoPut some space between the groups of tests
Pierre Ossman [Wed, 9 Jul 2014 12:12:39 +0000 (14:12 +0200)]
Put some space between the groups of tests

10 years agoAdd more 8 bpp conversion tests
Pierre Ossman [Wed, 9 Jul 2014 12:12:12 +0000 (14:12 +0200)]
Add more 8 bpp conversion tests

10 years agoOnly grab the regions we actually need
Pierre Ossman [Thu, 13 Feb 2014 08:42:48 +0000 (09:42 +0100)]
Only grab the regions we actually need

Allows pixel buffers that do damage tracking to get
proper information.

10 years agoProvide a better R/W base PixelBuffer class
Pierre Ossman [Thu, 6 Feb 2014 15:31:10 +0000 (16:31 +0100)]
Provide a better R/W base PixelBuffer class

Clearly separates the read API from the write API
and also from actual implementation.

10 years agoGet rid of getStride() 9/head
Pierre Ossman [Thu, 30 Jan 2014 16:57:27 +0000 (17:57 +0100)]
Get rid of getStride()

It was confusing and not properly used everywhere.
Callers should use the stride they get when they get
the buffer pointer.

10 years agoGet rid of the direct access abuse of FullFramePixelBuffer's data
Pierre Ossman [Thu, 30 Jan 2014 16:47:31 +0000 (17:47 +0100)]
Get rid of the direct access abuse of FullFramePixelBuffer's data

10 years agoWe don't need a fillRect() that is this optimised
Pierre Ossman [Thu, 30 Jan 2014 15:59:14 +0000 (16:59 +0100)]
We don't need a fillRect() that is this optimised

Keep things simple instead and allows us to remove the extra setPF()
methods.

10 years agoAvoid having virtual methods where not needed
Pierre Ossman [Tue, 28 Jan 2014 12:47:18 +0000 (13:47 +0100)]
Avoid having virtual methods where not needed

10 years agogetImage() doesn't need to be overloadable, keep it simple
Pierre Ossman [Wed, 12 Feb 2014 12:23:30 +0000 (13:23 +0100)]
getImage() doesn't need to be overloadable, keep it simple

10 years agoCreate a proper interface base class for the viewport's framebuffer
Pierre Ossman [Fri, 7 Feb 2014 13:46:26 +0000 (14:46 +0100)]
Create a proper interface base class for the viewport's framebuffer

This allows us to gracefully fall back to the FLTK code in case the
platform specific code cannot be used.

10 years agoEncoders/decoders should track the connection object
Pierre Ossman [Fri, 31 Jan 2014 11:37:32 +0000 (12:37 +0100)]
Encoders/decoders should track the connection object

The connection object is a much more appropriate object for the
decoders and encoders to keep track of. Besides the streams, it also
contains state like connection parameters.

10 years agoPush encoder and decoder handling down into the connection objects
Pierre Ossman [Fri, 31 Jan 2014 12:12:18 +0000 (13:12 +0100)]
Push encoder and decoder handling down into the connection objects

This keeps the reader and writer objects clean and simple protocol
decoders/encoders.

10 years agoRemove full support for colour maps
Pierre Ossman [Mon, 20 Jan 2014 14:05:21 +0000 (15:05 +0100)]
Remove full support for colour maps

Gets rid of a loooot of code and complexity.

Colour map clients are still supported through an
automatically generated map, but we lose the ability to
develop a client or server that uses colour maps
internally.

10 years agoGet rid of unnecessary macros as they are only ever set to a single thing
Pierre Ossman [Wed, 12 Feb 2014 12:16:43 +0000 (13:16 +0100)]
Get rid of unnecessary macros as they are only ever set to a single thing

10 years agoPush encoding specific formats into the encoders and decoders
Pierre Ossman [Mon, 17 Mar 2014 13:35:51 +0000 (14:35 +0100)]
Push encoding specific formats into the encoders and decoders

Keep the generic stream classes clean and general.

10 years agoConsolidate the different palette handler implementations
Pierre Ossman [Fri, 7 Mar 2014 12:48:29 +0000 (13:48 +0100)]
Consolidate the different palette handler implementations

10 years agoHandle CopyRect like any other encoding
Pierre Ossman [Fri, 31 Jan 2014 10:21:51 +0000 (11:21 +0100)]
Handle CopyRect like any other encoding

Avoids having to special case things. Keeps the code simpler.

10 years agoAvoid reader/writer duplication in every encoder/decoder
Pierre Ossman [Wed, 29 Jan 2014 15:42:48 +0000 (16:42 +0100)]
Avoid reader/writer duplication in every encoder/decoder

10 years agoRemove partial update functionality
Pierre Ossman [Tue, 21 Jan 2014 13:45:10 +0000 (14:45 +0100)]
Remove partial update functionality

Get rid of the ability of encoders to only send part of the
rectangle they are given as we don't want partial updates
anyway.

10 years agoMove serialisation of pseudorects into separate functions
Pierre Ossman [Thu, 16 Jan 2014 12:39:05 +0000 (13:39 +0100)]
Move serialisation of pseudorects into separate functions

Keeps the code clearer.

10 years agoMerge the "V3" message classes into the normal ones
Pierre Ossman [Thu, 16 Jan 2014 12:12:40 +0000 (13:12 +0100)]
Merge the "V3" message classes into the normal ones

We have no need for this abstraction so let's keep things simple.

10 years agoMerge branch 'pixeltest' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 7 Jul 2014 12:36:35 +0000 (14:36 +0200)]
Merge branch 'pixeltest' of https://github.com/CendioOssman/tigervnc

10 years agoMerge branch 'cleanup' of https://github.com/CendioOssman/tigervnc
Pierre Ossman [Mon, 7 Jul 2014 12:27:04 +0000 (14:27 +0200)]
Merge branch 'cleanup' of https://github.com/CendioOssman/tigervnc

10 years agoRemove a lot of platform compatibilty stuff 3/head
Pierre Ossman [Mon, 7 Jul 2014 12:13:46 +0000 (14:13 +0200)]
Remove a lot of platform compatibilty stuff

It's either not used, or no longer relevant.

10 years agoRemove remnants of Windows only client
Pierre Ossman [Mon, 7 Jul 2014 11:44:03 +0000 (13:44 +0200)]
Remove remnants of Windows only client

10 years agoFix some offenders that poke around in the PixelFormat internals
Pierre Ossman [Wed, 12 Feb 2014 12:12:31 +0000 (13:12 +0100)]
Fix some offenders that poke around in the PixelFormat internals

10 years agoAvoid code duplication for the linear pixel format conversion methods
Pierre Ossman [Thu, 30 Jan 2014 08:51:39 +0000 (09:51 +0100)]
Avoid code duplication for the linear pixel format conversion methods

10 years agoConsistent use of stride vs pitch
Pierre Ossman [Wed, 22 Jan 2014 10:28:05 +0000 (11:28 +0100)]
Consistent use of stride vs pitch

Consistently use the term stride rather than pitch. Also
consistently represent the stride in number of pixels rather
than number of bytes. There is so much code that assumes
proper alignment already that we do not need the extra resolution.

10 years agoFix possible information leakage
Pierre Ossman [Mon, 20 Jan 2014 16:11:19 +0000 (17:11 +0100)]
Fix possible information leakage

Zero the padding byte used when converting 24-bit RGB to 32-bit.

10 years agoUse the same algorithm for conversions
Pierre Ossman [Mon, 20 Jan 2014 15:40:10 +0000 (16:40 +0100)]
Use the same algorithm for conversions

Make sure that the routines that convert between buffer
and rgb representations follow the same algorithm so it's
easier to compare them.

10 years agoOptimise the RGB to pixel conversion to simple shifts
Pierre Ossman [Mon, 20 Jan 2014 16:23:51 +0000 (17:23 +0100)]
Optimise the RGB to pixel conversion to simple shifts

Might as well make it inline at this point as well as it is used
heavily in other routines that convert entire blocks of pixels.