answer is truncated and the resolver switches to TCP, causing the pollfd
descriptor will change. Any poll on this old descriptor will either
timeout or continually poll as ready
SVN revision: 71776
enhance keyboard input handling.
* Associate the keymap with the input device rather than the display
since you could could have different keymaps associated with different
devices.
* Increase the size of character arrays used for the string
representations of
the keyname, keysym and for the string representing the key.
* Re-enable the code that converts the keysym to a printable
definition - this
is required where the keysym is not the same as the printable definition
SVN revision: 71750
& modifier change(s). Also, add patches from Rob Bradford
<rob@linux.intel.com> for basic support for setting wayland cursor:
From c5d9094f4792bf7375c216682947de334002cdad Mon Sep 17 00:00:00 2001
From: Rob Bradford <rob@linux.intel.com>
Date: Tue, 29 May 2012 16:35:01 +0100
Subject: [PATCH] Ecore_Wayland: Add basic support for setting the cursor
* Add api to the ecore_wl_input_ namespace to allow setting the buffer
to use
for the pointer and for loading a named cursor from a cursor theme.
Under the
Wayland protocol the cursor is associated with the input device.
* Add helper functions to ecore_wl_window to set the cursor based on the
active pointer input device for the window.
* Load the cursor theme when the SHM interface is ready and provide an
API
call to provide a wl_cursor for a given name.
* Add API to restore to the default cursor and then use that when the
pointer
enters the surface to ensure compliance with the Wayland protocol.
Cheers :)
SVN revision: 71734
dns.c is the work of William Ahern (http://25thandclement.com/~william/projects/dns.c.html) and has been in development for ~4 years.
it has zero documentation, no examples, and no comments: I'm pretty sure he's an EFL developer in disguise.
this new resolver is roughly 10-12% faster than c-ares, will never randomly break, requires no external libraries, and is much simpler on the efl side.
by default, dns.c resolution is enabled any time ipv6 support is detected and c-ares support is not requested;
it should work on all platforms (wink wink vtorri), but it has no way of disabling ipv6 (someone can go through dns.c/h and do that if they feel so inclined)
SVN revision: 71690
ecore_x is calling XKeycodeToKeysym from xlib, but this function has issues. It
has been deprecated 6 months ago in X.org and it is suggested to use
XkbKeycodeToKeysym instead: https://bugs.freedesktop.org/show_bug.cgi?id=25732
This commit makes a wrapper function that makes use of XkbKeycodeToKeysym when
possible or fallback to XKeycodeToKeysym otherwise.
SVN revision: 71677
NOTE: VIRTUAL part are almost like rectangle except they don't create any object
on the canvas. This part can't be visible, nor have any color, nor be used as a
clip, nor receive any event.
SVN revision: 71674
This is a list that holds all requests being processed at the moment. It
can be used so that when a new request is added, first we look in this
queue to avoid creating a new request that is the same as the one being
processed right now. If it is already being processed, a new waiter is
added to it, just like when the request was on its original queue.
SVN revision: 71607
slave.
It will now receive errors from the thread callback and call the error
request function, with the respective error type in the message.
SVN revision: 71606
The implementation of the slave doesn't need to care about reading
commands and sending answers. It just receives the arguments for its job
and returns the processed data.
SVN revision: 71599
pthread_create can fail if to much pthread are pending. As pthread_join
is only called when the main loop can flush the async call list, there was
a possibility for edje_cc to run out of thread due to that. We know
force the flush of the async call list and so call pthread_join when
pthread_create fail.
SVN revision: 71555
Subject: [E-devel] Ecore: Save proper evas engine buffer depth when
_ecore_evas_resize is done
I have found bug in ecore evas buffer. It is created by default as RGB32,
and then it can be changed by ecore_evas_alpha_set function, but when resize
is done it is always created as ARGB32.
Patch in attachment checks in resize if alpha channel is set, and then set
appropriate buffer depth.
SVN revision: 71537
Also, put the slant angle calculations in a macro for easier future changes.
Just have it there so people who want it can turn it on.
SVN revision: 71506
A #link at the beginning of a new line goes interpreted by doxygen as a title,
so format the documentation to avoid this issue. No content change.
SVN revision: 71501
The challenge here is that the native window representation is stored
in Ecore_Evas's prop.window. But currently there is no checking of
what driver the Ecore_Evas is for when calls are made to e.g.
ecore_evas_software_x11_window_get.
The attached change to Ecore makes the appropriate functions return 0
or NULL if the driver for the Ecore doesn't match as expected. This
can then be used to identify if an Ecore_Evas is e.g. from X11 or from
Wayland.
SVN revision: 71453
index is already used in string.h, avoid it here.
src/modules/immodules/xim/ecore_imf_xim.c:116: warning: declaration of 'index' shadows a global declaration
/usr/include/string.h:487: warning: shadowed declaration is here
Signed-off-by: Stefan Schmidt <s.schmidt@samsung.com>
SVN revision: 71442
Subject: Re: [E-devel] [Patch] Add Ecore_X_Error_Code enumeration
I added the Ecore_X_Error_Code enumeration which wraps X error codes.
I think this will be useful when the X error occurs.
SVN revision: 71379
The Slave_Proc now inherits from Slave, which implements all the
communication logic. The Slave_Proc only has specific code for
processes, while a new Slave_Thread should be added soon with code for
slave threads.
SVN revision: 71368
Some efreet APIs do not check input parameters. So I add checking by
using EINA_SAFETY_ON_XXX().
ISO/IEC statndards says that "If an argument to a function has an
invalid value, behavior is undefined" . But this is just for the
primitive functions such as libc. I think that parameter checking is
needed in at least EFL exported APIs to prevent run-time abnormal
behavior.
EINA_SAFTETY_ON_XXX are better than "if (xxx) return" because it gives
error message and can be maintainable.
Patch by Bluezery, modified by me
SVN revision: 71366
Make it possible to do it, and make it default.
And remove the now useless valgrind option (we want to see still
reachable now that libcheck works).
SVN revision: 71316
Subject: [E-devel] [Edje]: Bug Fix: Edje draggable jumps when external
events is used.
Please find attached bug fix patch for edje draggable jump issue when
external event area is used.
Bug: When an external event area is used for edje draggable and when
after mouse move if immediate mouse down
is done then the draggable jumps back to its original position.
Analysis: In _edje_mouse_down_signal_cb When an external event area
is set i.e., when rp->events_to is set.
tmp.x value is set to 0, need_reset is set to 1 and also
_edje_recalc_do is called including emitting "drag" signal. this code
is
unnecessary/buggy and instead it causes the jump.
1. In mouse down only drag->down.x and drag->down.y needs to be set
which is being set below and tmp value need not be reset to
0 as tmp value is calculated in mouse move based on drag->down.x and
drag->down.y values.
2. need_reset is already set in mouse up hence need not be set in
mouse down again.
3. edje_recalc_do is the function which actually causes the movement
of draggable based on tmp value hence need not be called in mouse down.
because of the above code race condition happens and as tmp value is
being set to 0 and need reset is also enabled the draggable jumps back
to where it
started.
4. "drag": is sent even before "drag,start" [ should not /need not be
sent in mouse down ]
All the above code is added only when external event area is set and
the above code is not even related to whether external event is set or
not.
Solution: When an external event area is set directly equating rp =
rp->events_to and sending mouse,down would be enough, as down.x and
down.y is set below
including sending drag,start. Recalc_do should be called only in mouse
move as its responsible for movement including setting tmp value.
need_reset is already set in mouse up. drag should not be sent from
mouse down.
Change Description:
Bug Fix: Edje Draggable jumps when mouse down is done immediately
after mouse move when an external
event area is used.
demo edc pasted below to reproduce the issue.
Please find attached bug fix patch for edje draggable jump issue when external event area is used.
Bug: When an external event area is used for edje draggable and when after mouse move if immediate mouse down
is done then the draggable jumps back to its original position.
Analysis: In _edje_mouse_down_signal_cb When an external event area is set i.e., when rp->events_to is set.
tmp.x value is set to 0, need_reset is set to 1 and also _edje_recalc_do is called including emitting "drag" signal. this code is
unnecessary/buggy and instead it causes the jump.
1. In mouse down only drag->down.x and drag->down.y needs to be set which is being set below and tmp value need not be reset to
0 as tmp value is calculated in mouse move based on drag->down.x and drag->down.y values.
2. need_reset is already set in mouse up hence need not be set in mouse down again.
3. edje_recalc_do is the function which actually causes the movement of draggable based on tmp value hence need not be called in mouse down.
because of the above code race condition happens and as tmp value is being set to 0 and need reset is also enabled the draggable jumps back to where it
started.
4. "drag": is sent even before "drag,start" [ should not /need not be sent in mouse down ]
All the above code is added only when external event area is set and the above code is not even related to whether external event is set or not.
Solution: When an external event area is set directly equating rp = rp->events_to and sending mouse,down would be enough, as down.x and down.y is set below
including sending drag,start. Recalc_do should be called only in mouse move as its responsible for movement including setting tmp value. need_reset is already set in mouse up. drag should not be sent from mouse down.
Change Description:
Bug Fix: Edje Draggable jumps when mouse down is done immediately after mouse move when an external
event area is used.
demo edc pasted below to reproduce the issue.
Please find attached bug fix patch for edje draggable jump issue when
external event area is used.
Bug: When an external event area is used for edje draggable and when
after mouse move if immediate mouse down
is done then the draggable jumps back to its original position.
Analysis: In _edje_mouse_down_signal_cb When an external event area
is set i.e., when rp->events_to is set.
tmp.x value is set to 0, need_reset is set to 1 and also
_edje_recalc_do is called including emitting "drag" signal. this code
is
unnecessary/buggy and instead it causes the jump.
1. In mouse down only drag->down.x and drag->down.y needs to be set
which is being set below and tmp value need not be reset to
0 as tmp value is calculated in mouse move based on drag->down.x and
drag->down.y values.
2. need_reset is already set in mouse up hence need not be set in
mouse down again.
3. edje_recalc_do is the function which actually causes the movement
of draggable based on tmp value hence need not be called in mouse down.
because of the above code race condition happens and as tmp value is
being set to 0 and need reset is also enabled the draggable jumps back
to where it
started.
4. "drag": is sent even before "drag,start" [ should not /need not be
sent in mouse down ]
All the above code is added only when external event area is set and
the above code is not even related to whether external event is set or
not.
Solution: When an external event area is set directly equating rp =
rp->events_to and sending mouse,down would be enough, as down.x and
down.y is set below
including sending drag,start. Recalc_do should be called only in mouse
move as its responsible for movement including setting tmp value.
need_reset is already set in mouse up. drag should not be sent from
mouse down.
Change Description:
Bug Fix: Edje Draggable jumps when mouse down is done immediately
after mouse move when an external
event area is used.
demo edc pasted below to reproduce the issue.
Please find attached bug fix patch for edje draggable jump issue when external event area is used.
Bug: When an external event area is used for edje draggable and when after mouse move if immediate mouse down
is done then the draggable jumps back to its original position.
Analysis: In _edje_mouse_down_signal_cb When an external event area is set i.e., when rp->events_to is set.
tmp.x value is set to 0, need_reset is set to 1 and also _edje_recalc_do is called including emitting "drag" signal. this code is
unnecessary/buggy and instead it causes the jump.
1. In mouse down only drag->down.x and drag->down.y needs to be set which is being set below and tmp value need not be reset to
0 as tmp value is calculated in mouse move based on drag->down.x and drag->down.y values.
2. need_reset is already set in mouse up hence need not be set in mouse down again.
3. edje_recalc_do is the function which actually causes the movement of draggable based on tmp value hence need not be called in mouse down.
because of the above code race condition happens and as tmp value is being set to 0 and need reset is also enabled the draggable jumps back to where it
started.
4. "drag": is sent even before "drag,start" [ should not /need not be sent in mouse down ]
All the above code is added only when external event area is set and the above code is not even related to whether external event is set or not.
Solution: When an external event area is set directly equating rp = rp->events_to and sending mouse,down would be enough, as down.x and down.y is set below
including sending drag,start. Recalc_do should be called only in mouse move as its responsible for movement including setting tmp value. need_reset is already set in mouse up. drag should not be sent from mouse down.
Change Description:
Bug Fix: Edje Draggable jumps when mouse down is done immediately after mouse move when an external
event area is used.
SVN revision: 71277
NOTE: know issue, in elementary_config the size of the icon
change after a theme reload. I don't know what information is
lost between to reload. If someone can point at them, thanks.
SVN revision: 71235
NOTE: as librsvg is a massive source of bugs in e17, it is now
removed from evas. You can still use librsvg by using the
evas_generic_loader. Please not that you need to properly delete
it from your disk if you don't use a package manager. The file to
remove :
/*/lib/evas/modules/loaders/svg/linux-gnu-i686-1.2.*/module.so
SVN revision: 71223
Now evas will in all case do the layout during the prepare stage. It will do that
once and as long as the text didn't change. This does improve by a factor of at
least 2.3 in all expedite test case except the text change that only get a 30%
increase (I expect a drop in performance on non pipe rendering for text change
expedite test only, but this case is not common in real life).
This also fix the issue that show random size glyph when using pipe rendering.
SVN revision: 71220
There is no automatic promotion of unsigned to unsigned long when using va_arg,
which means it is illegal to pass an 'unsigned' value and then use it as an
unsigned long in eina_arg_vset. Doing so yields incorrect results on some
architectures like itanium
Patch by Albin 'Lutin' Tonnerre <albin.tonnerre@gmail.com>
SVN revision: 71196
The eina_value code TYPE_CHAR conversion code assumes that 'char' is a signed
type, which is not true on some platforms like ARM and PPC. We need to
explicitely use signed chars to make sure the value is correct.
Patch by Albin 'Lutin' Tonnerre <albin.tonnerre@gmail.com>
SVN revision: 71195
supports FreeBSD >= 7 these days), so the check for __FreeBSD_version >=
420001 is not necessary anymore (plus it probably never worked, as that
macro is defined in sys/param.h, which is not included prior to the
check).
Patch by Raphael Kubo da Costa
SVN revision: 71172
glyph metrics.
Instead of having to render the glyph to get the width and horizontal
bearing of it, it's possible to get this information from the glyph
metrics (which are available on the glyph slot).
This change now allows Evas to only render the glyph at the rendering
phase, instead of having to render it during layout phase.
SVN revision: 71132
(-fastcomp and -fastdecomp) -fastcomp makes for faster decompressing
AND faster compressing of edj files, -fastdecomp is a bit slower on
compression but also as fast as -fastcomp in decompression. note that
edje files built with these optiosn will not work on older edje
installations, thus they are options.
SVN revision: 71112
in wayland compositors. Added alpha support for wayland_egl. Support
evas output rotation in wayland_egl. Don't move/resize windows in
wayland_egl unless sizes actually change. Included patch from Robert
Bradford <robert.bradford@intel.com> for vertical/horizontal mouse
wheel scrolling.
SVN revision: 71108
This commit also includes patch(s) from Robert Bradford
<robert.bradford@intel.com> for Supporting vertical/horizontal
scrolling, and updates to wayland fixed point for input events.
Fix ecore_wl_input to use new libxkbcommon api.
Add new surface_enter/leave listener for ecore_wl_window.
SVN revision: 71107
This is in preperation of a future change to be able to set errors in
function calls as well, and not just constructors.
Also, I improved the error reporting.
SVN revision: 71000
Currently, this feature is only supported in EGL/GLESv2 environment
with GL_IMG_multisampled_render_to_texture extension supported.
_____________________
from: (sanghee park) sh15.park@samsung.com
Dear all,
I compose this mail to ask reviewal this patch about multisampling on the evasgl.
I want to make multisampling capacity to enhance rendering quality of the evasgl.
But if MSAA is applied always, this have possibility lowering rendering performance,
I separated user's input level to high, mid, low, none.
If you want to test this patch, try to examine rendering qulity on EGL circumstance with multisampling level.
Plaese review it, and any suggestion will be appreciated.
Best Regards,
SangHee
SVN revision: 70992
Subject: [E-devel] [PATCH][EDJE] Patch to remove the alpha from image
header while saving if the alpha is set to 1 but the image is fully
opaque
Attached to the mail is a patch to set the alpha information for an
image header to 0 with alpha present but all the texels being opaque.
Continuing to our discussion, as suggested by many people in the
community it has been implemented at edje_cc level.
Change description:
While compiling the edc file, image data for image files is
scanned to find out whether the alpha value in header is set to 1 and
is not being used in the image.
If this is the case, while writing to eet the alpha is set to 0 to
avoid blending for such images in the graphics pipeline when used by
evas.
SVN revision: 70954
- Cleanup cache2 things on shutdown
- Use Eina_File instead of straight shm_open + mmap when loading things from cserve2
- Do free the mapped images when we don't need them
SVN revision: 70936
As we move back and forth from double to fixed point, we do
have some rounding error. I am trying to limit them at much as
possible by reducing the number of computation in double.
SVN revision: 70905
NOTE: now you can change theme dynamically in elementary apps more reliably.
This doesn't handle the case where the swallow was done in a parent object and
the reswallow should happen in a another group. I don't how to fix that use
case.
don't see yet how to handle that
SVN revision: 70901
For the moment only edje_player use it. This means that when used with
edje_watch, you don't need any more to type any kind of command line
when you are testing value in your theme. As a side effect, this means
that their is a real use case to make edje_cc faster !
SVN revision: 70890
edje_watch call edje_cc and monitor all the source file (edc, font
image, sound). If any of them change, it call edje_cc, update its
watching list and so on. edje_watch as the same command line as
edje_cc.
Still a little bit rought, but it's the beginning of an interesting
experiment.
SVN revision: 70872
Mainly, glDeleteBuffers was being called instead of glDeleteRenderbuffers.
Also, there was an error when checking if surface is valid.
SVN revision: 70870
Hello.
Just 3 small fixes to get our warning count down. The tempget one
should actually save us against wrong reads.
Also a small DSO fix reported and confirmed in IRC.
Please review and apply.
regards
Stefan Schmidt
Submitted-By-Off: Stefan Schmidt<stefan@datenfreihafen.org>
SVN revision: 70759
now seriously...
Introducing Cache Serve 2.
This cache server will initially load images for clients connected to
it. It starts slave processes to load these images, and share the loaded
images through shm with the clients. All the connection done between
clients and the server goes through sockets.
The cserve2 build option is turned on by default, while the old cserve
was disabled, but in order to make clients use it, the environment
variable EVAS_CSERVE2 must be set, and a server must be running.
Clients will try to find the socket on a specified location using the
environment variable EVAS_CSERVE2_SOCKET. If it's not defined, then the
XDG_RUNTIME_DIR path should be used, and finally HOME, TMPDIR and /tmp.
SVN revision: 70699
(Trying it again since this commit broke evas build yesterday.)
Previously, evas_gl_surface_create() didn't actually do
the render buffer attach to the the FBO. It was performed when
the make_current was called for the first time. The issue
was that even though the surface was successfully created with
the given configuration, there was a possibility of make_current
failing with the error message "FBO not complete" because of
the surface configuration.
So, I've added a piece of code that checks the FBO
capabilities beforehand to set up a available surface configurations
so that it doesn't have to fail during make_current for unsupported
surface format.
Also, I've changed the surface config in a way that once the
user calls evas_gl_surface_create(), evas gl sets the config
parameter with configuration that evas_gl is actually using.
SVN revision: 70680
Complete support for keyboard events. Thank You :)
NB: This is a modified patch from what Rob originally sent. This fixes
formatting, uint32_t types, function name, and other such things ;)
SVN revision: 70672
Previously, evas_gl_surface_create() didn't actually do
the render buffer attach to the the FBO. It was performed when
the make_current was called for the first time. The issue
was that even though the surface was successfully created with
the given configuration, there was a possibility of make_current
failing with the error message "FBO not complete" because of
the surface configuration.
So, I've added a piece of code that checks the FBO
capabilities beforehand to set up a available surface configurations
so that it doesn't have to fail during make_current for unsupported
surface format.
Also, I've changed the surface config in a way that once the
user calls evas_gl_surface_create(), evas gl sets the config
parameter with configuration that evas_gl is actually using.
SVN revision: 70617
This make it possible to use the object tree to reduce the number of object, we need to explore to know
what is under a specific position. First used by propagation event code. That code is now 4 times faster, enjoy !
As a side cost evas_object_move goes from 925 to 980 valgrind cycle on my computer, so not something you
will notice.
NOTE: if you notice any breakage regarding event propagation, map, cats, minor or major, please report to
me ! I hope I didn't loose my mojo, with such a scary change, I have a big chance to get it back !
SVN revision: 70564
Subject: [E-devel] [patch] missing doxygen files in release tarballs
This patch add to EXTRA_DIST essential files for doxygen
small build fix:
SVN revision: 70514
NOTE: This should be part of evas_render itself and not
delegated to the engine. So cleaning things to make it easier
during evas_render rewrite.
SVN revision: 70503
Note: this two were broken. Current plan to bring
that feature back in, is to attach this information
to Evas_Text_Prop during the prepare stage. This
would improve both single and multiple core rendering
without increasing the number of lock and the complexity
of the code.
SVN revision: 70501
NOTE: other things that may join it in the near futur EVAS_FRAME_QUEUE,
EVAS_METRIC_CACHE and maybe EVAS_WORD_CACHE also. This is all part of
cleaning up our rendering path so we can actually improve it more easily.
SVN revision: 70499
clients not moving or resizing properly with most recent git wayland.
Make use of wayland's new 'serial' stuff in place of timestamps (where
appropriate). Add code to handle new wayland 'ping' events.
SVN revision: 70443
eobj_event_callback_del -> eobj_event_callback_del_lazy.
eobj_event_callback_del_full -> eobj_event_callback_del.
Thanks to cedric for the suggestion.
SVN revision: 70435
It seems that libcheck keeps some reachable data, unfortunate as it makes
it very annoying for us to check for reachable memory in our code, but
letting valgrind report about it is just too damn annoying.
SVN revision: 70412
This macro makes the code a tad simpler, but more importantly, makes it
easier for us to be thread safe, or more corrctly, easier for us the
make user code thread safe.
SVN revision: 70407
Subject: [E-devel] [patch] eina_simple_xml example
Here is an example for eina_simple_xml
This patch includes a small sample XML file, source code (for parsing and
printing it out) and the doxygen doc.
SVN revision: 70385
NOTE: some of this function should be moved as inline, but that's to late for a change
I think. So we will fix that if needed.
Second point, I am not happy with is eina_inarray_insert and eina_inarray_insert_at. The
naming is really poor.
SVN revision: 70352