>
> In my testing (x86_64/gcc), this had no impact on the resulting binary,
> building with/without it gives the same stripped binary save for its
> buildid.
>
> Signed-off-by: Christophe Fergeau
Have you compiled proposed Snir plugin?
I think we are going to use soon
> On 28 Feb 2018, at 17:35, Christophe Fergeau wrote:
>
> s/statments/statements in the short log.
Fixed
>
> On Wed, Feb 28, 2018 at 04:43:06PM +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> This patch ensures that all
> On 28 Feb 2018, at 18:16, Frediano Ziglio wrote:
>
>>
>> From: Christophe de Dinechin
>>
>> Fix a race condition, make sure we only use stream after opening
>> Get rid of C-style 'goto done' in do_capture.
>> Get rid of global streamfd, pass it
> On 1 Mar 2018, at 10:51, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 17:36 +0100, Christophe Fergeau wrote:
>> My understanding is that the previous iteration was quite controversial,
>> I would just drop it from the series unless you get acks from everyone
>> involved
> On 28 Feb 2018, at 17:36, Christophe Fergeau wrote:
>
> My understanding is that the previous iteration was quite controversial,
> I would just drop it from the series unless you get acks from everyone
> involved this time.
It’s a bit difficult to drop that from the
> On 1 Mar 2018, at 11:02, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Use C++ "placement new" to directly initialize a C++ object at a
>> specified memory location.
> On 1 Mar 2018, at 11:59, Frediano Ziglio wrote:
>
>>
>> From: Christophe de Dinechin
>>
>> The 'Stream' class is designed to abstract file I/O. In a subsequent
>> patch, message formatting will be isolated out of the class, but in
>> order to
> On 1 Mar 2018, at 11:56, Frediano Ziglio wrote:
>
>>
>> From: Christophe de Dinechin
>>
>> It makes sense to place the codecs inside the Stream which is charged
>> with setting it up, until we create a proper abstraction for input
>> messages
> On 1 Mar 2018, at 11:51, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> It makes sense to place the codecs inside the Stream which is charged
>> with setting it up,
> On 1 Mar 2018, at 14:45, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> - The Stream class now deals with locking and sending messages
>> - The Message<> template class
> On 1 Mar 2018, at 14:56, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> There are two classes:
>> - The X11CursorUpdater class sends the messages. It can be used when
>>
> On 1 Mar 2018, at 18:19, Jonathon Jongsma wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Until we have a proper separation of input messages, it makes sense
>> to
>> have the Stream
> On 1 Mar 2018, at 15:46, Lukáš Hrázký wrote:
>
> On Thu, 2018-03-01 at 12:40 +0100, Lukáš Hrázký wrote:
>> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>>> From: Christophe de Dinechin
>>>
>>> Throwing 'runtime_error' directly
> On 1 Mar 2018, at 13:13, Frediano Ziglio wrote:
>
>>
>> From: Christophe de Dinechin
>>
>> Throwing 'runtime_error' directly should be reserved for the support
>> library. Add an 'Error' class as a base class for all errors thrown
>> by the
On Thu, Mar 01, 2018 at 10:03:13AM -0600, Jonathon Jongsma wrote:
> On Thu, 2018-03-01 at 16:41 +0100, Christophe de Dinechin wrote:
> > > On 1 Mar 2018, at 16:27, Christophe Fergeau
> > > wrote:
> > >
> > > Signed-off-by: Christophe Fergeau
> > > ---
>
> On 1 Mar 2018, at 12:40, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Throwing 'runtime_error' directly should be reserved for the support
>> library. Add an 'Error'
Tests require 'catch' to be installed, one might want to disable them if
catch is not available. This patch adds a --disable-tests switch. By
default, tests are enabled depending on 'catch' availability.
Signed-off-by: Christophe Fergeau
---
configure.ac| 21
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Until we have a proper separation of input messages, it makes sense
> to
> have the Stream class deal with this flag.
>
> Signed-off-by: Christophe de Dinechin
On Thu, 2018-03-01 at 16:27 +0100, Christophe Fergeau wrote:
> The .desktop file is generated, so will be in $(builddir), not
> $(srcdir)
>
> Signed-off-by: Christophe Fergeau
> ---
> Makefile.am | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git
> On 28 Feb 2018, at 17:46, Christophe Fergeau wrote:
>
> On Wed, Feb 28, 2018 at 04:43:15PM +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Throwing 'runtime_error' directly should be reserved for the support
>> library.
> On 1 Mar 2018, at 15:54, Lukáš Hrázký wrote:
>
> On Thu, 2018-03-01 at 15:42 +0100, Christophe de Dinechin wrote:
>>> On 1 Mar 2018, at 15:11, Lukáš Hrázký wrote:
>>>
>>> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
From:
> On 1 Mar 2018, at 16:11, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Doing this change will make it possible to move the capture loop to the
>> concrete-agent.cpp
> On 1 Mar 2018, at 16:47, Lukáš Hrázký wrote:
>
> On Thu, 2018-03-01 at 16:23 +0100, Christophe de Dinechin wrote:
>>> On 1 Mar 2018, at 15:42, Lukáš Hrázký wrote:
>>>
>>> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
From:
> On 1 Mar 2018, at 17:02, Christophe de Dinechin
> wrote:
>
>
>
>> On 1 Mar 2018, at 16:12, Lukáš Hrázký wrote:
>>
>> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>>> From: Christophe de Dinechin
On Thu, 2018-03-01 at 16:41 +0100, Christophe de Dinechin wrote:
> > On 1 Mar 2018, at 16:27, Christophe Fergeau
> > wrote:
> >
> > Signed-off-by: Christophe Fergeau
> > ---
> > configure.ac | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> On 1 Mar 2018, at 16:12, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> This makes it possible to remove X11 dependencies from the main agent file.
>>
>> Note: it may
On Thu, 2018-03-01 at 16:27 +0100, Christophe Fergeau wrote:
> Tests are enabled by default, but are expensive to compile, so they can be
> disabled if that's what one wants. They will also be enabled/disabled by
> default depending on the availability of 'catch'
Thanks again for this :)
I'd
On Thu, 2018-03-01 at 16:23 +0100, Christophe de Dinechin wrote:
> > On 1 Mar 2018, at 15:42, Lukáš Hrázký wrote:
> >
> > On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> > > From: Christophe de Dinechin
> > >
> > > Appropriate error
> On 1 Mar 2018, at 16:14, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Isolating classes in separate files makes parallel builds faster,
>> facilitates code reuse and
> On 1 Mar 2018, at 16:27, Christophe Fergeau wrote:
>
> Convenience script to avoid running autoreconf/configure manually.
> It can be run from a directory out of the source tree for VPATH builds.
> This script is based off the one in spice-gtk.
>
> Signed-off-by:
> On 1 Mar 2018, at 16:27, Christophe Fergeau wrote:
>
> Upstream provides a .pc file, we can use it rather than doing the
> detection manually.
Acked-by: Christophe de Dinechin
Will test on macOS too, there are some specific aspects of libjpeg .pc
> On 1 Mar 2018, at 16:42, Christophe Fergeau wrote:
>
> Hey,
>
> I sent
> https://lists.freedesktop.org/archives/spice-devel/2018-March/042479.html
> (which is based on spice-gtk's autogen.sh) which should address the
> issues below.
Will ack there.
>
> Christophe
>
Hey,
I sent
https://lists.freedesktop.org/archives/spice-devel/2018-March/042479.html
(which is based on spice-gtk's autogen.sh) which should address the
issues below.
Christophe
On Fri, Feb 16, 2018 at 03:28:41PM +, Daniel P. Berrangé wrote:
> On Fri, Feb 16, 2018 at 04:18:19PM +0100,
> On 1 Mar 2018, at 16:27, Christophe Fergeau wrote:
>
> Signed-off-by: Christophe Fergeau
> ---
> configure.ac | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/configure.ac b/configure.ac
> index 2ab14da..ce83153 100644
> ---
> On 1 Mar 2018, at 16:27, Christophe Fergeau wrote:
>
> In my testing (x86_64/gcc), this had no impact on the resulting binary,
> building with/without it gives the same stripped binary save for its
> buildid.
I was also thinking that -fvisibility=hidden did not make
> On 13 Feb 2018, at 16:54, Frediano Ziglio wrote:
>
> Changes since v4:
> - split reset patch;
> - add a workaround for a Qemu bug, seems easier to do it
> instead of detecting it;
> - add another problem with fix and test case.
For what it’s worth, I tested this for a
Upstream provides a .pc file, we can use it rather than doing the
detection manually.
Signed-off-by: Christophe Fergeau
---
configure.ac | 14 +-
1 file changed, 1 insertion(+), 13 deletions(-)
diff --git a/configure.ac b/configure.ac
index 5aab662..e5decb3
The .desktop file is generated, so will be in $(builddir), not $(srcdir)
Signed-off-by: Christophe Fergeau
---
Makefile.am | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Makefile.am b/Makefile.am
index 74ec16a..94ad7aa 100644
--- a/Makefile.am
+++
Signed-off-by: Christophe Fergeau
---
configure.ac | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/configure.ac b/configure.ac
index 2ab14da..ce83153 100644
--- a/configure.ac
+++ b/configure.ac
@@ -50,7 +50,7 @@ LIBVIRT_LINKER_NO_INDIRECT
Convenience script to avoid running autoreconf/configure manually.
It can be run from a directory out of the source tree for VPATH builds.
This script is based off the one in spice-gtk.
Signed-off-by: Christophe Fergeau
---
autogen.sh | 18 ++
1 file
Hey, while looking at adding a --disable-tests argument as discussed
recently on the mailing list, I piled up a couple of unrelated changes
to configure.ac/Makefile.am.
Christophe
___
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
In my testing (x86_64/gcc), this had no impact on the resulting binary,
building with/without it gives the same stripped binary save for its
buildid.
Signed-off-by: Christophe Fergeau
---
configure.ac| 19 ---
src/Makefile.am | 2 --
2 files changed, 21
Tests are enabled by default, but are expensive to compile, so they can be
disabled if that's what one wants. They will also be enabled/disabled by
default depending on the availability of 'catch'
Signed-off-by: Christophe Fergeau
---
configure.ac| 21
> On 1 Mar 2018, at 15:42, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> Appropriate error classes been created for the existing messages:
>> - ProtocolError: handles the
On Thu, 2018-03-01 at 10:17 +, Frediano Ziglio wrote:
> Options are global and should be ignored if not recognised by a
> specific plugin as they can be used by other ones.
Haven't realized that..
Acked-by: Lukáš Hrázký
> Signed-off-by: Frediano Ziglio
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Isolating classes in separate files makes parallel builds faster,
> facilitates code reuse and minimizes the chances of patch conflicts.
>
> Signed-off-by: Christophe de
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> This makes it possible to remove X11 dependencies from the main agent file.
>
> Note: it may be unsafe to call XCloseDisplay from the destructor.
> Doing some experiments
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Doing this change will make it possible to move the capture loop to the
> concrete-agent.cpp file.
>
> Signed-off-by: Christophe de Dinechin
> ---
>
On Thu, 2018-03-01 at 15:42 +0100, Christophe de Dinechin wrote:
> > On 1 Mar 2018, at 15:11, Lukáš Hrázký wrote:
> >
> > On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> > > From: Christophe de Dinechin
> > >
> > > The FrameLog
On Thu, 2018-03-01 at 12:40 +0100, Lukáš Hrázký wrote:
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> > From: Christophe de Dinechin
> >
> > Throwing 'runtime_error' directly should be reserved for the support
> > library. Add an 'Error' class as a
> On 1 Mar 2018, at 15:11, Lukáš Hrázký wrote:
>
> On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
>> From: Christophe de Dinechin
>>
>> The FrameLog encapsulates the frame binary log, making sure that the
>> log is properly closed in
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Appropriate error classes been created for the existing messages:
> - ProtocolError: handles the various kinds of protocol-related errors
> - MessageDataError: a form of
On Thu, 2018-03-01 at 02:32 -0500, Frediano Ziglio wrote:
> >
> > On Wed, 2018-02-28 at 15:42 -0600, Jonathon Jongsma wrote:
> > > On Wed, 2018-02-28 at 15:34 -0600, Jonathon Jongsma wrote:
> > > > On Thu, 2018-02-22 at 11:40 -0500, Frediano Ziglio wrote:
> > > > > >
> > > > > > Gstreamer based
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> The FrameLog encapsulates the frame binary log, making sure that the
> log is properly closed in case of exceptions.
>
> Signed-off-by: Christophe de Dinechin
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> There are two classes:
> - The X11CursorUpdater class sends the messages. It can be used when
> no thread is required.
> - The X11CursorThread spawns an X11CursorUpdater in
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> - The Stream class now deals with locking and sending messages
> - The Message<> template class deals with the general writing mechanisms
> - Classes, FormatMessage,
>
> In some bad cases, for example, host OS crashes without
> sending any FIN to usbredirserver, and usbredirserver
> will keep idle connection for a long time.
>
> We can also set the kernel arguments, it means that other
> processes may be affected.
>
> Setting a sensible timeout(like 10
>
> From: Christophe de Dinechin
>
> Throwing 'runtime_error' directly should be reserved for the support
> library. Add an 'Error' class as a base class for all errors thrown
> by the streaming agent, as well as subclasses used to discriminate
> between categories of
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Throwing 'runtime_error' directly should be reserved for the support
> library. Add an 'Error' class as a base class for all errors thrown
> by the streaming agent, as well
>
> From: Christophe de Dinechin
>
> The 'Stream' class is designed to abstract file I/O. In a subsequent
> patch, message formatting will be isolated out of the class, but in
> order to minimize code changes, this intermediate step simply moves
> the corresponding
>
> From: Christophe de Dinechin
>
> It makes sense to place the codecs inside the Stream which is charged
> with setting it up, until we create a proper abstraction for input
> messages similar to what Message does for output messages in follow-up
> patches.
>
>
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> It makes sense to place the codecs inside the Stream which is charged
> with setting it up, until we create a proper abstraction for input
> messages similar to what Message
On Thu, Mar 01, 2018 at 11:10:55AM +0100, Jakub Janků wrote:
> From: Jakub Janků
>
> Add --with-gtk configure option.
> If used, favor GTK+ over Xlib.
Looks good to me both patches.
I'll push early next week if no one complains about it.
Acked-by: Victor Toso
On Wed, Feb 28, 2018 at 01:33:35PM +0100, Lukáš Hrázký wrote:
> On Wed, 2018-02-28 at 12:19 +0100, Christophe Fergeau wrote:
> > On Wed, Feb 28, 2018 at 11:56:07AM +0100, Lukáš Hrázký wrote:
> > > On Tue, 2018-02-27 at 18:02 +0100, Christophe Fergeau wrote:
> > > > On Tue, Feb 27, 2018 at
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> The 'Stream' class is designed to abstract file I/O. In a subsequent
> patch, message formatting will be isolated out of the class, but in
> order to minimize code changes,
Options are global and should be ignored if not recognised by a
specific plugin as they can be used by other ones.
Signed-off-by: Frediano Ziglio
---
src/mjpeg-fallback.cpp| 2 --
src/unittests/test-mjpeg-fallback.cpp | 7 +++
2 files changed, 3
From: Jakub Janků
Introduce new functions to handle clipboard,
add new files clipboard.[ch]
This is only a preparatory patch for
following GTK+ clipboard implementation.
---
Makefile.am | 2 ++
src/vdagent/clipboard.c | 68
From: Jakub Janků
---
src/vdagent/clipboard.c | 416 +++-
src/vdagent/clipboard.h | 6 +-
src/vdagent/vdagent.c | 23 ++-
src/vdagent/x11-priv.h | 44 ++---
src/vdagent/x11.c | 25 ++-
src/vdagent/x11.h | 6 +
From: Jakub Janků
Add --with-gtk configure option.
If used, favor GTK+ over Xlib.
---
configure.ac | 13 -
src/vdagent/vdagent.c | 8 +++-
src/vdagent/x11.c | 6 ++
3 files changed, 25 insertions(+), 2 deletions(-)
diff --git a/configure.ac
From: Jakub Janků
This code using Xlib was removed by commit
"x11: retrieve _NET_WM_NAME using GDK"
a4f5c7dcb6579f61434a443e537641f91da6cb8c.
Bring it back to make GTK+ optional for now.
---
src/vdagent/vdagent.c | 6 ++--
src/vdagent/x11.c | 85
On Wed, 2018-02-28 at 16:43 +0100, Christophe de Dinechin wrote:
> From: Christophe de Dinechin
>
> Use C++ "placement new" to directly initialize a C++ object at a
> specified memory location. This makes it possible to get rid of
> type-unsafe memset, low-level unsafe
On Wed, 2018-02-28 at 17:36 +0100, Christophe Fergeau wrote:
> My understanding is that the previous iteration was quite controversial,
> I would just drop it from the series unless you get acks from everyone
> involved this time.
I've only commented on compiler support, it seems it's fine, so no
In some bad cases, for example, host OS crashes without
sending any FIN to usbredirserver, and usbredirserver
will keep idle connection for a long time.
We can also set the kernel arguments, it means that other
processes may be affected.
Setting a sensible timeout(like 10 minutes) seems good.
From: Victor Toso
Cosmetic change. One break was dropped on _default:_ label of switch
Signed-off-by: Victor Toso
---
src/channel-usbredir.c | 8 +---
src/spice-channel.c| 7 +--
2 files changed, 10 insertions(+), 5 deletions(-)
diff
74 matches
Mail list logo