Summary: Support of Spinlocks in Eina (Eina_Spinlock) for OSX, which does not implement them in pthread.
@feature
Reviewers: raster, raoulh, naguirre, cedric, stefan_schmidt
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1151
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: This is the first step to introduce a gl-drm backend.
Test Plan: "ecore evas" create with ecore_evas_gl_drm_new(). It creates "ecore evas" with gl_drm evas backend.
@feature
Reviewers: raster, Hermet, cedric, devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1187
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
added support of all action types that use targets to
edje_edit_program_target_add and edje_edit_program_target_del
added targets cleaning to edje_edit_program_action_set to avoid usage of program id
as part id and vise versa.
@fix
Reviewers: cedric, seoz, Hermet
Subscribers: cedric, reutskiy.v.v
Differential Revision: https://phab.enlightenment.org/D1342
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This fixes broken rendering of the main surface when new surface is created.
Rendering on main surface can be broken due to invalid size of common gl context.
Since common gl context is shared between surfaces, if new surface has been changed
size of common gl context, then rendering on main surface can be broken.
In order to fix problem, engine has to change size of common gl context after egl
make current according to size of current Outbuf.
@fix
Test Plan:
run elementary_test with wayland_egl engine under wayland and click any
button to create new surface. watch the main surface of elementary_test.
Reviewers: devilhorns, raster, stefan_schmidt, cedric
Reviewed By: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1346
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
use eng_output_resize of software_generic engine and move wl_egl_window_resize to
eng_outbuf_reconfigure. and remove eng_output_resize of wayland_egl.
@fix
Test Plan: run elementary_test with wayland_egl engine under wayland
Reviewers: devilhorns, raster, stefan_schmidt, cedric
Reviewed By: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1343
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
On Ubuntu 14.04 it makes a 32 bit depth window un-responsive
to any XEvent.
Reviewers: cedric, raster
Reviewed By: raster
Subscribers: raster, capOM, cedric
Differential Revision: https://phab.enlightenment.org/D1236
Because of Eina_Value being less than optimal in our usage, we're dropping it in Eolian.
This simplifies the code and makes it easier to bind to other languages, which
will aid new generators. Also, we're dropping long double support from eo files
and expressions as it causes an ABI breakage in gcc 4.4.
This change adds a new API to retrieve the legacy prefix for enum members
and also adds support for generating enums into the C generator. Besides that,
it prevents generation of entire struct contents when the .eo file declares
a named struct inside of a typedef - as the struct itself gets generated later on.
I don't know what I was thinking, struct and enum fields need to be ordered
so now we keep a separate list around containing the field names in correct order.
This implements the previously declared APIs to manipulate enum types. Also,
it implements the appropriate lookups in expression evaluation so that you
can refer to constants and enums inside constant expressions.
Also added API to retrieve value of a variable (forgot about that one)
and made it impossible to create extern constants (doesn't make sense)
as well as made it impossible to give extern globals values.
Nothing parses yet, and no API is exposed. Also, move the remaining instances of file
inside of existing structures to Eolian_Object and parse basename only once.
This way we can only lex expr related tokens (operators etc.) when actually
about to parse an expression. That allows stuff like nested complex types
without the lexer treating the endings as right shift.
so instead of "unsigned int" you get "uint". This is important for handling of
expressions and cross-language interoperability. You can use c_type_get on the
base type to get the C name. Also, append the appropriate suffix to number literals
when calling eolian_expression_value_to_literal.
This commit also does several side (related) changes. Particularly, it updates
the Eolian C generator to use the new API, it adds missing expr types (null, char)
and masks, updates the API dealing with default return values to use expressions
instead of strings and does several fixes (mainly in lexer) around the place.
It also disallows single quoted strings as those are reserved for characters.
For this purpose, several other changes were made. There can now only be 1 class per file
and the class name has to match the file name. Also, I replaced the list of classes with
a hash, allowing for fast indexing (crucial for parser performance - otherwise it was
very slow).
This also adds a new API, eolian_type_class_get.
This tests that even with multiline ellipsis, the ellipsis in the second
line is not overdone (i.e it only does ellipsis as needed).
This is a test for something we found out when we applied D1311 (for T1213).
This patch fixes an issue causing text to be cut off in some cases.
The problem was that we were calculating line width and alignment before
we did any bidi calculations, which in turn caused us to use the wrong
text items for those calculations.
Many thanks to Daniel Hirt for investigating this deeply, finding all
the nitty-gritty and generally pointing me to where the problem is.
Daniel also provided the test case.
His patch (D1291) was close, but not enough.
Fixes T1496
@Fix
Summary:
If none of ECORE_IMF_MODULE and WAYLAND_DISPLAY variables are setted
best of available moduls should be found. Removed odd return statement.
Reviewers: cedric, seoz, Hermet
Reviewed By: Hermet
Subscribers: cedric, reutskiy.v.v
Differential Revision: https://phab.enlightenment.org/D1344
@fix
This reverts commit d408408283.
this breaks mult-line "long" filenames in efm. 2nd line is just ...
for almost all of them (ones that are actually in need of 3 or more
lines). break break! REVERT!
Summary:
Evas Textblock ellipsis is handled in a item.
When the ellipsis item is added in the text, some characters are cut off
considering width of ellipsis character.
But, it is handled in only one text item.
If there are many short text item, the ellipsis item can be cut off visually.
Fixes Phab ticket T1213
@fix
Test Plan: This commit includes test case.
Reviewers: woohyun, seoz, sohyun, tasn
Subscribers: herdsman, cedric
Differential Revision: https://phab.enlightenment.org/D1311
Reset the Outbuf's Engine information pointer.
We also don't need to do an eng_window_free during reconfigure because the
software_generic_update function will free it anyway.
We should also re-setup the tilebuffer during eng_setup.
Fix gl_context resize to work even if there is no wl_egl window setup
yet.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This is the first part of the wayland-egl engine fix. Don't resize the
gl_context during the first_rectangle function, and don't set any
preserve bits on the gl_context
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
If we fail to init gl_generic then we should cleanup properly and free
the recently created Outbuf.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
there are deep down bugs in evas with animated gifs. espeically if you
have multiple instances of the same gif, but for now let's just
address the problem where we have entire frames of animation
vanishing. this is because the animation frame set failed and thus
didnt notify the rest of evas. have it ignore this fail for now so
things work out.
the text object walked all text items and took the biggest. when that
big item used some fallback font that didnt match the primary font in
ascent/descent etc. things went wrong - text expanded when not
expected to. this makes text objects just use the primary font
ascent/descent metrics and fixes e titlebar magically expanding thus
compressing window content when not expected.
This fix T1459. I have added an ERR to catch this kind of issue earlier.
I am wondering if that should not be even a CRI.
The reason why we do see the problem only after the introduction of the
use of Eina_Rectangle_Pool is due to how efficiently they pack data, resulting
in ressource ending up in the wrong format bucket. Nothing wrong with the
patch itself.
so the test suites were useless because they didn't ecompile what was
generated. it was missing a ) at the end of calling eo_do_super to
start with. it was missing a call to do_uper fo destructors too. also
it didnt deal with the #include "blah.eo.c" at the file end either so
it produced uncompilable src at the start. again - also fix. this also
fixes test case comparisons to match the now correct generation that
compiles.
Summary:
@fix
Segfault in wayland_egl engine is casused by illegal library linking.
Fix this by linking to GLESv2 and EGL libraries.
Test Plan: N/A
Reviewers: devilhorns, raster, cedric, Hermet
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1332
several loaders make too much noise when a file fails to load - almost
all of this is because the file isnt actually an image format file at
all, and an ap is just asking evas to try do a load to see if it is a
loadable image at all. this results in noisy strerr output that simply
shouldnt be there. so silence for loaders. @fix
i found that we are not setting u and v to be smooth (linear
interpolate) for yuv reendering, even when asked. they remain at a
default "nearest". this enables linear for u and v always, meaning
even when smooth is off, we still interpolate u and v (not y), and
even when at 1:1 with no scaling u and v get interpolation for better
quality.
@fix!
this fixes the elipsis changes xmike made by assuming elipsis of 0
whenn text min x is set (which is what we want anyway). this is a
complaint that is all through phab and other places where people
complain about "x..." or something now replacing formerly perfectly
find labels
The reversal makes it possible to merge two #if and unspaghetti the code
a bit.
Since the diff is not very readable, here is the before/after to show
the spirit:
before:
#if cond
if (...)
#endif
do_foo();
#if cond
else
do_bar();
#endif
after:
#if cond
if ( ! ...)
do_bar();
else
#endif
do_foo();
this fixes CID 1039647 - it caught a real bug. we were not skipping
out on an empty + or - push/pop format fomamnt if it had nothing
there. not actually something people do - but possible. fix!
there is a possible deadlock condition where locks are not taken in
the same order (img lock then cancel lock). re-order so this doesn't
happen. this fixes CID 1106339
since colors are hex, 96 bytes of hex would be 32bytes per r, g and b,
and that would be 128 bits per r g and b and that just is never seen
or used. also coverity is right in CID 1193201 that we just can't
shift that many, so go down to max 16 bits per rgb which is sensible
and able to be done. (thats 12 bytes max for hex of color).
[Problem] When obj which has src object(proxy concept) are excluded from render_object,
src object's 'proxy.redraw' isn't change although src obj have not proxies.
because 'proxy.redraw' flag is changed to EINA_FALSE only in obj's subrender() called.
[Resolution] When the count of source obj's proxies is 0, src obj's 'proxy.redraw' is set for EINA_FALSE.
Signed-Off-By: Min Kyoung Kim <mer.kim@samsung.com>
Summary:
When the grid is mirrored, originally code is just changed (x1,y1) to (x2,y2).
I fix calculate logic for the x-axis.
@fix
Test Plan:
1. elementary_test -> grid
2. click UI-MIRRORING toggle
Reviewers: Hermet, woohyun, raster, seoz
Reviewed By: seoz
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1325
Summary:
wrong structure, a Edje_Image_Directory_Set_Entry, used for copy
a set's data to new file.
@fix
Reviewers: raster, seoz, cedric, Hermet
Reviewed By: Hermet
Subscribers: Sachiel, cedric, seoz
Differential Revision: https://phab.enlightenment.org/D1304
We cannot use the texture to find a valid format Before the texture is
actually created. This is invalid, leads to "warning: tex is used
uninitialized' message from the compiler, and just plain wrong.
Instead, use the alpha property of the Evas_GL_Image to find the
proper texture format, Then create the texture.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
We use conn here for a global Dbus connection variable as well as sometimes
in a local scope. Better rename the global one to avoid confusion and
potential failures when the code would fallback to the global one even if
the local scope should be used and the var was simply forgotten.
i hope this addresses CID 1229131 - don't trust the DISPLAY var
content much at all - limit it to [a-z][A-Z][0-9][-] only. hopefully
coverity is happier.
coverity pointed out a deadlock - CID 1226193. i hope this fixes it
(need to wait days for a coverity re-run). tests pass and real life
usage works, so i assume its ok. in theory i see no problem either.
in looking at CID 1230994 i noticed that eina_unicode_unicode_to_utf8
and eina_unicode_utf8_to_unicode are really dumb. they calloc an array
of bytes then proceed to fill them all in anyway. why? also the
realloc handing in eina_unicode_unicode_to_utf8 wasn't strictly corect
and could leak memory. so this just fixes silly slow code and a leak.
this addresses CID 1230994. as such eina_unicode_unicode_to_utf8()
always returns a nul terminated string. so it's guaranteed. but yes -
if string is 7 bytes or longer it will not put a nul byte on the
destination. as such for a single unicode char this can never happen
as in utf8 it's 6 bytes. but since eina_unicode_unicode_to_utf8()
safely returns a nul terminated string at all times - we can just use
strcpy safely. no need for strncpy. also handle null return from
eina_unicode_unicode_to_utf8()
so in fixing one CID for a format lookup failing and thus invalid mem
access, a possible leak was made in this case. fix that by doing
lookup before any allocation and if lookup fails, return there
this fixes CID 1230999 1230998 1230997
this fixes potential leak due to
ff652084e0 fixing a valgrind access
issue. make sure the server is in the list before handling error
conditions in case a callback sets delete_me (should never though).
gl_generic functions expect the Outbuf to contain the canvas, so we
fixed the eng_window_new function to accept that, And the engine info.
This simplifies the calls to eng_window_new.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Since we simplified the eng_window_new function to accept the canvas
and engine info, we need to update calls to that function
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
We need to pass in the canvas and engine info structure as the
gl_generic engine needs the outbuf->canvas to be set.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary:
Evas wayland_egl engine crashes on startup when it is running with mesa.
Evas is trying to get EGL / GLES extension function addresses where evas_gl_symbols.
It is called before eglGetDisplay. Then mesa considers that current egl platform is
X11 if it has no EGL_PLATFORM env var. This behavior of mesa is bad. But we should
provide workaround until it is fixed. Thus EGL_PLATFORM env var should be defined to
wayland before calling eglGetProcAddress.
@fix
Test Plan: run elementary_test under wayland with wayland_egl engine.
Reviewers: raster, stefan_schmidt, devilhorns
Reviewed By: devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1312
this should fix T1508 - my testing seems to indicate so. test suite
succeeds, some manual tests of mine work. so is ay this is a good fix
that could fix critical segvs in apps using ecore-con - eg like
weather module in e
scim has a nasty habit of sitting and hanging if tryng to set up and
no scim daemon is there. since in x11 you may have multiple input
methods, leave it to be manually set up via env var, and if not then
dont work at all.
this fixes ecore-imf to auto-pick scim or wayland vased in DISPLAY and
WALAND_DISPLAY env vars. it's not a perfect/good/correct fix. it
should be picking based on canvas engine type, but that is a far more
invasive effort, so for release, do this simple one that brings back
wayland input panel and keeps x11 input methods going as long as app
is either x11 (DISPLAY) or wayland (WAYLAND_DISPLAY). if both are set,
it'll get it wrong. as i said - that's a more invasive fix.
Seems the newer wayland-egl code was missing the log domain variable
and thus would not load at time of dlsym due to missing symbol. This
fixes that problem by defining the variable.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
th pending cancel may call callbacks... which can mess with the list
and thus the pending_add we have, sso do our work before calling this
(like set pending_call to NULL).
Also fixed some incorrect @context
Changed the @edcsubsection tags so they live in first line of the doccomment, this way you can read them when the file is folded-closed, improving the code readability.