I'll reintroduce this in a moment, was just easier this way in order to
merge Vincent's changes.
This reverts commit d46fa601ef45d579d1448b19887d02ff2dfbb576.
Summary:
Label size is incorrectly calculated (text does not wrap) if
elm_label_line_wrap_set() is called before elm_label_wrap_width_set().
@fix
Test Plan:
The label3 test case is added to show the bug case.
In label3 test case, if we switch the calling sequence of
elm_label_line_wrap_set() and elm_label_wrap_width_set(),
the result is different.
Reviewers: raster, seoz, woohyun, Hermet
Differential Revision: https://phab.enlightenment.org/D1369
Summary: Removed all global variables for item class
Reviewers: zmike, raster, seoz
Maniphest Tasks: T1482
Differential Revision: https://phab.enlightenment.org/D1318
Summary:
If gengrid select mode is not ELM_OBJECT_SELECT_MODE_ALWAYS, ELM_OBJECT_SELECT_MODE_ALWAYS mode for
elm_genlist_item_select_mode was broken. In the item select routine, item mode(it->mode) was not checked.
So, I added the check routine there.
@fix
Test Plan: I revised the elementary_test code and tested with that.
Reviewers: raster, seoz
Reviewed By: seoz
Subscribers: SanghyeonLee, bluezery
Differential Revision: https://phab.enlightenment.org/D1372
This completely breaks automatic UI mirroring. The code there was
correct. Please, before removing code, contact the original author and
make sure you don't break things down. Especially if your reason for
removing is "this doesn't make sense to me" instead of "this breaks X
Y."
This reverts commit 24983fed239a3b37c05cc3c65ed9c56953a7176b.
thre are still all the old engines supported inside, but docs dont say
so so effectively deprecated. also handle a NULL engine. this is part
of our engine/display system cleanup.
Summary:
elm_widget calls _item_del_pre_hook of each widget and widget can
decide item deletion by return type. But gengrid and popup widget implemented
those _item_del_pre_hook function as a void type.
So this commit fixs the return types to Eina_Bool and changes gengrid widget to return EINA_FALSE when
the item wants to be deleted later.
And when the item wants to be deleted directly, it returns EINA_TRUE.
@fix
Test Plan: test in demo.
Reviewers: seoz, raster, bluezery
Subscribers: Hermet, jaehwan
Differential Revision: https://phab.enlightenment.org/D872
now that we have ELM_DISPLAY and ELM_ACCEL we have little need for a
lot of the elm engine stuff and frankly a lot was broken if you used
multiple engines etc. so this fixes a lot of it nd removes a lot of
cruft. i think i got it all right, but i may have oppsied. this also
remo9ves engine support for engines long gone from evas.
Summary:
ELM_OBJECT_SELECT_MODE_ALWAYS mode for elm_genlist_item_select_mode
was broken. In the item select routine, item mode(it->mode) was not
checked. So, I added the check routine there.
@fix
Test Plan: I revised the elementary_test code and tested with that.
Reviewers: raster, seoz
Reviewed By: seoz
Subscribers: SanghyeonLee, bluezery
Differential Revision: https://phab.enlightenment.org/D1347
This reverts commit 2e2d003efeded27955cc9301279cd3a0d362f548.
Revert this patch to leave the credit to Andrii Kroitor.
I will fix this correctly in the next commit.
Sorry Andrii for the wrong review comments that can break compatibility.
But actually it was not documented at all so there was no problem.
Just to please hermet, let's not do more.
This reverts commit a2afbe91e851f3d943736d2a1aae9e2721d1a0d5.
definitely this will break compatibility.
and as i mentioned, we should prevent as possible that
until there are other methods.
Summary:
added support of bottom_left, bottom_right, top_left, top_right styles
of notify that are present in default theme
@fix
Reviewers: cedric, seoz, Hermet
Subscribers: reutskiy.v.v
Differential Revision: https://phab.enlightenment.org/D1329
this makes the black rect etc. appear instantly even if no effects are
selected for comp - this should cover up lockscreen and other stuff on
suspend/shutdown/resume etc. like fading does. this should fix T1498
if entry has selection, it would be cleared if selection region set is requested again.
because entry cancel the selection area after it reset the selection area.
unfortunately, the selection area is new selection so selection area is gone.
this patch prevent clearing selection and keep the new selection.
@fix
Now both hoversel and button are usable.
This is easiest way to fix the bug. But why it didn't work well even
hoversel inherit button default style?
Revert "button: revert this patch because this patch just broke hoversel."
This reverts commit 3238c8787511aaf69b755760b24f1aa518477d75.
hoversel is now usable. let's fix the button warning message with more
considerations.
Revert "button: add fixed 1 1 in case that text is not used."
This reverts commit efd33ddb5ba51dccaee2734e8a7add18801d9285.
Draggable area could be zero size if user set the panes content size
with elm_panes_xxx_size_set().
In this case, it prints fixed messages.
now, it's gone and i tested with various cases for any side effects.
the engine checks are also broken - but that's a much bigger change,
so this fixes elm config and main code to not do their x11 work if no
DISPLAY env var is set. this is right for a wayland world where you
don't want DISPLAY but WAYLAND_DISPLAY set and used. this fixes
several ERRs that should never have hapapned in this setup.
lots of code unsing ENGINE_COMPARE where it should get the engine from
the ecore_evas directly - it happened to work as we happened to have
apps that have windows only for x11, or wayland etc, but not both in
an app. this was wrong, so remove the engine compare macro and do it
properly. no laziness.
this was yet another thing broken by the mostly-unreviewed and untested widget item focus feature. it's still not completely fixed, since the reorder modes are broken, but at least it works for the general case again
@fix
@fix
moved the user callbacks call at the end of the function, so the user
is able to modify the list without making the code below the call to
fail miserably.
Also improved the Genlist Del test to also include this case.
@fix
To calc the position in a table you need to make the modulo of the col/row size, not the modulo on the x/y value
This unbreak the elm_gengrid_item_pos_get() function that was reporting wrong values
Especially the -fvisibility=hidden would be needed here for our EAPI handling.
I never realised we miss this here but Lukasz Stanislawski pointed me to it
when we wondered why it linked ofr him with missing EAPI but not for me.
I'v compiling and releasing elm like this for a long time now so I don't
expect any trouble but if something strange comes up we can still revert it
before 1.11 final.
Issue reported by Stefan Schmidt. Above atspi function had EAPI attribute
defined in implementation file, however this symbol should not be public in
elementary 1.11.
These headers are installed in the system but not officially supported
as they are considered as internal headers.
The reason why they still need to be installed is that
- they were exposed even in the previous releases.
- they are needed for widget customization (unofficially)
- or it will break ewe.
Special thanks to Stefan for the reporting.
Summary:
Elementary application crashes on startup when it is running with
wayland engine. This crash problem happens while ecore_x_screensaver API
is being called from _win_noblank_eval funcion. Thus we should not call
ecore_x_screensaver APIs if elm win doesn't have X11 window.
Test Plan:
1. build efl and elementary to support x11 and also wayland
1. run weston without xwayland
2. run elementary_test on the weston
Reviewers: raster, seoz, stefan_schmidt, devilhorns
Differential Revision: https://phab.enlightenment.org/D1293
misspelled.
In case someone exports the ELM_ENGINE to be wayland-shm or
wayland-egl, let's handle That case also. Typically it is set to
wayland_shm or wayland_egl but since we check variants of spelling for
other engines also, then let's check it for the wayland engines too.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
As a small fallout from e7dbe063c9c044f072bacafd31ab872f6210788b we now
need to set ELM_DISPLAY with our system has something like $DISPLAY or
$WAYLAND_DISPLAY set. Elm uses that now to find out which window system
to use.
In the case of daing make doc we want it forced to shot even if we run
under X. With ELM_DISPLAY we do that. Thanks goes to raster for the
explanantion and git bisect for pinning it down.
In 540741dd48d0720876a5e68b2aa4aa7a9d79b8a3 one screenshot was
moved into HAVE_CXX_11 guards to allow building with cxx11 disabled.
But in the case of cxx11 enabled we want all screenshots plus this
one so append here instead of redefine.
this deals with an issue brought up today on irc - seemingly rage
asking for accel is causing segvs in vbox with its gl accel subsystem
- the only way to disable this is an env var, and this is unfriendly,
so make config gui and stuff to mess with accel preference to fix this.
Summary:
In case, when ctxpopup content have larger size than parent object, (by height, width or both) was resized popup at the full width or height of parent object size.
@fix
Reviewers: raster, seoz, cedric, Hermet
Reviewed By: Hermet
Differential Revision: https://phab.enlightenment.org/D1266
The -shared option in LDFLAG somehow is poisoning the configure run inside
distcheck. This results in the compiler test getting a segfault and failing.
I would really like to understand why this is needed to be set here manually
anyway. Same for fPIC and dPIC. I would expect that we leave this to the user
compiling the code or the packager. Skimming through the log only brought up
fixing c++11 globally. Not really a commit message that helps me to
understand why this is needed...
Summary: Do not set access info for naviframe prev_btn and next_btn when their access info are already set.
Reviewers: Hermet, kimcinoo
Reviewed By: kimcinoo
Differential Revision: https://phab.enlightenment.org/D1255
The Makefile.am in examples weren't conditioning the compilation of
C++ examples when C++ is disabled in configuration of elementary.
Which made the compilation to fail when --disable-c++11 is
used (in EFL or elementary) on C++ examples.
The example didn't include "elm_interface_atspi_widget_action.h" and
so failed on headers that used the type Elm_Atspi_Action in the
function's prototypes.
Summary:
Main purpose of exposing widget actions and keyboard shortcuts
is to allow accessibility clients to implement alternative methods
of GUI navigation.
Reviewers: z.kosinski
Reviewed By: z.kosinski
Subscribers: seoz
Differential Revision: https://phab.enlightenment.org/D1227
Don't call ecore_x functions unless we are running under an
X11-compatible engine, else we get segfaults in elm_apps.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
better - why?
1. no reliance on fnmatrch headers - have special enums for this so
fnmatch is an internal detail (casefole may not exist)
2. don't leak strduped strings - free them when done
3. have the same code for genlist and grid (dup for now until an
interface makes it the same search interface)
4. improve docs
5. get right @since version
6. use label get func in item class - providing a func won't work when
multiple items of multiple classes exist in the list
As discussed in IRC, the commit that added text_part_set,
text_part_get functions to elm_button are wrong. The correct functions
to be used are from elm_layout: text_set and text_get.
Fixed the example to use these functions that are inherited on
elm_button from elm_layout and fixed the cleanup of childrens
in the parent window by using the del event callback.
Summary: This function allows user to search for item in Gengrid.
Test Plan: elementary_test -to "Gengrid Item Search By Text"
Reviewers: seoz, singh.amitesh, Hermet, raster
Differential Revision: https://phab.enlightenment.org/D1178
- _ev_handler can't know whether elm_win still exists, so changed log style from notify to printf.
In addition, release the callbacks when elm_win is destroyed.
- use elm_sys_notify_send instead of elm_sys_notify_simple_send to show more functionality.
- remove static variables not to print warnings when with more than two test_sys_notify pannel.
Previously I removed -fPIC from example LDFLAGS and CXXFLAGS. Now
I'm adding "-fPIC -DPIC" to configure.ac conditioned to HAVE_CXX11.
This commit fixes 'make check'.
Summary:
This patch is dependent on D1193 and D1136.
It will be pushed after D1193 and D1136 patch.
Reviewers: singh.amitesh
Differential Revision: https://phab.enlightenment.org/D1168
Conflicts:
src/lib/elm_genlist.c
Summary: When item looping feature is on and press up or down key, screen seems to be frozen.
Test Plan: elementary_test -to "genlist focus" -> click item looing enable -> move list up and down
Reviewers: anand.km, seoz, woohyun
Subscribers: singh.amitesh
Differential Revision: https://phab.enlightenment.org/D1193
elm_test_clock.c:15:23: warning: declaration of ‘clock’ shadows a global declaration [-Wshadow]
elm_test_index.c:15:23: warning: declaration of ‘index’ shadows a global declaration [-Wshadow]
Removed portability warnings from Automake.
Removed *.eo.hh from BUILT_SOURCES.
Fixed some redeclaration of variables.
Added datarootdir and datadir to elementary.pc.in.
The legacy functions do unref after construction of the object. Which
doesn't happen on the eo_add calls directly. So it is needed to do
two unref's before letting the parent own the child.
C++ compilers refuse to compile typedefs for types not declared yet.
The typedefs in the elm_interface_atspi_accessible.h were out of order
with the definition (and consequently its declaration) of its struct.
This fixes by modifying the order of the typedefs to be after the
struct definition.