Hello all,
In elm_index.c, _item_del_pre_hook calls _item_free, where it->letter
is not set to NULL after eina_stringshare_del.
This will cause invalid access later on. The attached patch fixes the
problem, which is reproducible in my code before the fix.
Please check if this makes sense.
I am sorry that I was unable come up with a simple test app to
demonstrate the problem.
Thanks in advance.
Brian
Signed-Off-By:Brian Wang<brian.wang.0721@gmail.com>
SVN revision: 75723
Hi,
Attachment to the mail is elm_segment_control patch.
Change Log:
1. Generic Text support to segment items.
2. Generic Content support to segment items.
Signed-Off-By: PRINCE KUMAR DUBEY <prince.dubey>@samsung.com>
Thank you,
Prince
SVN revision: 75719
the wayland engines).
The focus handler which traps key events needs to listen for
ISO_Left_Tab also (which is what xkb reports for Shift+Tab).
SVN revision: 75618
NB: This is done because of the way slider handles the popup (indicator).
It is not getting trapped in the evas functions (or rather, the popup
falls through the traps there) so framespace is not accounted on the
popup, resulting in improper popup position when run in wayland.
NB: This is the only widget known to need this.
SVN revision: 75546
evas_object_resize call.
Not sure if this was a fat-finger c/p issue, or just plain
blindness/stupidity. How this remained in code for so long
without anybody seeing it is beyond me.
SVN revision: 75544
impl that manages copy & paste within a single process... this is an
example of how to implement cnp for another system and as well covers
fb cnp.. so terminology in fb has cnp work now :)
SVN revision: 75533
Subject: [E-devel] [Patch] elc_multibuttonentry resize patch.
Currently, there is a problem in the logic of getting the size in the
MBE's
box.
For each line, the height of the line is calculated according to the
biggest
item's height
and is applied to the box size, but problem occurs when the newly added
item's height is larger than the one before.
For example, when the entry is located in the front of the last line,
the line's height is determined by the entry.
When the button is added by the user's input at this state,
because button's height is larger than the entry's height
I have modified each line to have the equal height.
SVN revision: 75453
Dear, all
I've attached a patch that fixed the memory leak.
There is not a free() in _del_hook().
Please review this patch.
Sincerely yours,
Jihyeon Seol
Signed-Off-By: JiHyeon Seol<jiheyon.seol@samsung.com>
SVN revision: 75439
Subject: [E-devel] [patch][elementary] * access *
Series of of pathes from kim shinwoo. looked good to me - so in they
go, finishing off some more access mode to be more complete.
SVN revision: 75415
- swallow: do not use magic constant '12' and assume it was prefixed
with 'elm.swallow.';
- text: use part name and now always 'text'
SVN revision: 75338
the case of resetting the wayland cursor image. Fixes ticket #1293.
Increase the size of the event rects on the border theme to allow for
easier resizing.
SVN revision: 75313
Basically, we will remove the frame when going into fullscreen, and
readd when we leave fullscreen. When we remove the frame, then during
window redraw the appropriate fullscreen size will be calculated.
This allows proper fullscreening.
SVN revision: 75294
set it all the time themselves for read-only use outside. :) if uw ant
this u have to accomplish this another way. eg table with genlist PLUS
invisible rect obj at the same cell spot (0, 0, 1x1). and make the
invisible rect have a min size hint. table takes care of using the
"constraints" of both rect and gnelist then for sizing. :)
SVN revision: 75044
next/prev item - needed to add a feature to scrolelr (and accidentally
also to scrolled interface too :)) to make this work. now works.
SVN revision: 74969
1. on shutdown it calls lots of selected callbacks as it delets items
- wrong.
2. segv waiting to hapen due to list management (brian wang report).
SVN revision: 74956
Delete requests where not being called when the frame close button was
pressed. I know this is not the 'ideal' fix (which is why I left
myself a FIXME here), HOWEVER the Ideal fix would mean breaking the
Feature Freeze (and potentially API), so I will wait until freeze is
over for that. For now, this fixes the 'bugger' in a non-instrusive way.
SVN revision: 74946
resizing.
Previously, we only changed the mouse cursor when the border frame was
actually clicked. Now we can change it when the mouse moves over the
border/frame areas (like in X11). This adds a callback from the edc to
know when and where the mouse moves (with respect to the frame border
only). (IE: When the user moves the mouse over the bottom portion of
the border, the edc will let us know and we can change pointer
accordingly).
SVN revision: 74925
and "focus-out"."
This reverts commit 1bb12050b9785d47e102b4a3633210316d408f0d.
Conflicts:
ChangeLog
NEWS
Cedric, this is just wrong. The *on_focus* smart function of the base
widget is now the central place to handle focus in/out events.
I'll fix the spinner in sequence.
SVN revision: 74853
Widgets having internal panning objects won't have so fancy inheriting
schemas for those very objects, but they should be fine with
re-defining their own, if they want/need and overwriting it in their
smart_add()s.
SVN revision: 74682
Subject: [E-devel] [patch][elementary] ctxpopup, focus-chain
i'd like to raise an issue about focus chain on the ctxpopup. the focus
would go around only in the ctxpopup. (IMHO, raster keep the previous
popup
patch. so definitely!!)
but.. first of all, current ctxpopup does not have focus_next(). even
though there is focus_next(), it is not work properly,
because the parent of ctxpopup would be a widget which does not take
care
of sub objects in the focus_next(); such as elm_list.
(every widget can be a parent of ctxpopup, if the widget is enough
size to
display ctxpopup. thus the elm_list can be a parent of ctxpopup also).
anyhow, to resolve this issue, the patch uses event() of
elm_widget_event_hook_set().
the ctxpopup item is equal to the elm_list item. but the focus chain of
elm_list does not work properly.
generally, we use 'Tab' key to move focus, but elm_list uses the
(up,down)
arrow key moreover it is not a focus.
if you want to check this patch, it would be better to use
ELM_ACCESS_MODE=1.
SVN revision: 74650
Subject: [E-devel] [patch][elementary] dayselector - focus chain, and
about accessibility policy.
dayselector does not work in focus cycle because there is no
focus_chain().
because dayselctor is a child of Elm_Layout_Smart_Class, so it is
enough to
use focus_chain() of layout.
and.. very carefully..
i'd like to raise an issue that currently visually impaired person
should
be obliged to click(tap) quickly two times to activate widget such as
button, check.
the implementation to check 2nd click in elm_access is great. but one
single click is enough to activate button or change state of elm_check.
moreover some widgets like dayselector which has customized style for
elm_check does not work properly, because they do something at signal
'mouse,down,1'.
SVN revision: 74649
Subject: [E-devel] [patch][elementary] naviframe - focus chain, access
i have attached a patch to fix focus chin issue when the item title is not
visible. and to add accessibility feature.
SVN revision: 74646
NOTE: This improve calendar performance quite a lot, but a better improvement
would be to populate the object when we really need it. I don't know how to
do that in current Elementary source code. Reason for another expected massive
improvement is that every time you set a property on an Elm_Calendar trigger
a populate and that one is costly (Around 6 populates call per Elm_Calendar in
our code example, meaning doing it could give us a 6 times boost...).
SVN revision: 74553
Subject: [E-devel] [patch][elementary] popup - focus chain,
accessibility
you would be in rush because of freezing.. but there is still hot. -_-;
i'd like to raise an issue about focus chain on the popup. The focus would
go around only in the popup. (IMHO)
but current popup gives a chance to get the focus to other widget which is
not in the popup.
please check the attachment which has accessibility feature also and give
your opinion. thanks.
SVN revision: 74467
initializing text parts first. Patch by Bluezery <ohpowel@gmail.com>
On Thu, Jul 26, 2012 at 2:29 PM, Bluezery <ohpowel@gmail.com> wrote:
> Hello,
>
> This patch make genlist free their text parts when each item is
unrealized.
> Before this patch, the item is just cached when an item is unrealized
> and the text part is freed and set by other text when the cached item
> is realized.
> This improves scrolling performance with very long text because this
> can reduce needless edje size recalculation time.
> Please review this patch.
>
SVN revision: 74413
This fixes "elm - genlist decorate all mode on and scroll - garbage
items left around."
Note: Do not manipulate swallowed items. Unswallow it first then
manipulate it.
SVN revision: 74412
Subject: [E-devel] [patch][elementary] popup - getting proper geometry value]
current popup does not give proper geometry value,
evas_object_geometry_get(popup, &x, &y, &w, &h) gives 0, 0, 0, 0.
the patch resolves it, but it would be not a good way to get geometry
value. which is the best way to set obj->cur.geometry.x(y,w,h)?
please enlighten me. thanks.
SVN revision: 74409
- fixed theme to look less like an arse:
- overlap theme is translucent
- naviframe title visibility toggle is now animated
- item pop/push animation now applies to title area as well
- icon without title label is now centered (see Naviframe 2 test)
- title label and subtitle are now centered among the free space, should not overlap buttons anymore.
- title will use a font-size range, trying to not overlap buttons anymore.
- title will use ellipsis, trying to not overlap buttons anymore.
- prev/next buttons do not have a fixed min/max size anymore
- emit signals elm,state,title_label,show and elm,state,title_label,hide
- by default the previous button label is the title of the previous page
however by using a regular elm_button with an icon and long label
exposes a bug with button theme. And there is a conceptual issue: if
the previous button label is too big, it will look horrible. We should
have a maximum size that, if reached, should have ellipsis. This
should be a different elm_button style, with
elm/button/base/naviframe/back_btn/default being an alias to it.
SVN revision: 74403
back button now uses standard icon "arrow_left", and provides a "Back"
text that should be translated with the application.
Then we can define the elm/button/base/naviframe/back_btn/default as
an alias to the default button. If a theme wants it could still
provide an alternative button. IOW no breaks.
SVN revision: 74399
In real world it's very likely you're flipping to something, back or
front, not randomly flipping around based on one button.
I'd say we should deprecate the elm_flip_go().
SVN revision: 74360
whenever we "go" flip while flipping, we'll enter an inconsitent
state. We should detect where to flip and flip back if required.
Test: go to "Flip" test, click "1" button twice, before the first
animation ends.
SVN revision: 74359
Please review this patch I don't know the intention why did someone add this flags on mouse up in scroller.
Anyhow I tested and works fine.
SVN revision: 74334
Subject: [E-devel] [PATCH] elm_icon is disappeared when theme hook is
called.
[Current Issue]
- The elm_icon can be disappeared when theme hook is called.
You can see the problem in the below situation.
a) elementary_test -> Layout select
b) elementary_config -> Fonts(toolbar) -> select some font
class, font,
style, size
c) Select "Apply" button
then two icons in title layout are disappeared.
[Main cause]
- when theme hook is called, internally
_elm_image_smart_sizing_eval
function is called.
The function calculates icon's min, max size.
But min, max size is calculated only in case no_scale is true or
resize_down or resize_up is false.
If application isn't set no_scale or resize_down/up, minw and
minh value
is just -1.
So when theme hooks is called then sizing_eval is called, icon's
min size
is -1 and that is disappeared.
[Change Description]
- I just added evas_object_size_hint_min_get(obj, &minw, minh) in
_elm_image_smart_sizing_eval.
Patch is working well, but I don't think this is right solution
because
that would break image(icon) min,max concept
SVN revision: 74163
Allows setting a trap in elm_win that intercepts calls to
ecore_evas. If there is a trap and the trap returns EINA_FALSE, then
the corresponding call is NOT issued. If it does not exist or returns
EINA_TRUE, then the call is executed.
Enlightenment window manager will set these traps and will call
e_border directly, allowing E17 to use Elementary! A major feature
given e_widgets painful usage.
This should also help integrating into Wayland or even debug.
SVN revision: 74156
widgets.
Some of them have initting code using the parent ptr for some logic.
Now there's a new idiom on instantiating widgets which adresses
that. It'll be used for all widgets from now on.
SVN revision: 74147
When smart obj was set as the contents the _configure would be called recursively.
In this process the lastest size could be reverted as the previous one.
SVN revision: 74064
Subject: [E-devel] [PATCH] Add frame size when calculating minimized
elm_win size
Hi,
I found frame size including width and height isn't counted in
_elm_win_resize_objects_eval() when calculating minimized elm_window
size.
It is OK for X engine because elementary only draw client area and X
provides widow frame. So both the width and height from
evas_output_framespace_get are 0.
But it cause bug for wayland engine because elementary need draw
window
frame by itself. So real client area size is smaller than window size.
If frame size isn't counted into minimized window size, there isn't
enough client area to layout widgets.
So it is bug for any engine in which elementary draws window frame by
itself. It is the reason of
http://trac.enlightenment.org/e/ticket/1064.
Could you please my attached patch for this issue?
Thanks.
SVN revision: 74049
The ERRs thrown on the terminal while exiting some tests were already
there -- check that those decorate and stuff modes add lots of NULL
children on lists.
Genlist work exactly as before. All the bugs and feature were kept :)
SVN revision: 73941
knew him well. disable because it seems nothing i try requests it
except thunderbird, and thunderbird seems to sumarrily ignore the text
we provide. why - don't know. would have to now dig into thunderbir
src, but suffice to say plaintext is working a charmer, so let's stick
to that.
SVN revision: 73913
elementary configuration environment. Patch by Bluezery
<ohpowel@gmail.com>
On Thu, Jul 12, 2012 at 1:00 AM, Bluezery <ohpowel@gmail.com> wrote:
> Hello,
>
> Vertical bounce enable value can be changed by elementary
> configuration environment.
> But currently It was always EINA_FALSE. I fix that this value can be
> got from elm_config value.
>
> Please review this patch.
> Thanks
SVN revision: 73596
ecore_evas_window_get with a replacement function that checks the
currently used engine first. This fixes a segfault when running elm
with the wayland engines.
SVN revision: 73568
We had an inconsistency of two missing getters, that would "leak" API
deprecation on specialized scrollers (scrollables). Putting them in
for good now.
SVN revision: 73540
i know - freeze is on. last thing going into elm. only because i
rejected the patch at the last minute and it was small enough to do
now.
SVN revision: 73450
Subject: [E-devel] [patch][elementary] diskselector, duplicate icon
for rounded item
minor fix to diskselector to repeat icons right.
The diskselector has round mode which can be set by
elm_diskselector_round_enabled_set().
In case of round mode is enabled, the icon does not display properly
because diskselector replace the icon for the additional items - not copy
and paste. To enable round mode, diskselector adds over/under itmes(additional
itmes) by using normal mode items. So there is a icon to set two places in
diskselector. But diskselector does not duplicate the icon.. this makes item
without icon but there should be.. I'm not sure the best way to duplicate icon.
SVN revision: 73397
Subject: Re: [E-devel] [Patch][layout] select layout in access mode
I revised the patch which has the way to get the state of
accessibility now. elm_layout_edje_object_can_access_set/get().
SVN revision: 73393
Its API will be the the common one on ALL scrollable widgets, too.
Any scrollable widget having a function behaviour clashing with one of
the elm_scroller_* will be marked as DEPRECATED from now on.
SVN revision: 73309
This is meant to be used for all widgets implementing scrolling views.
An Elementary interface adds a given behaviour to whichever widget,
from any class.
SVN revision: 73307
last to be evaluated.
Porting of the basic scrolling infra to an interface will happenfirst, as well as the scrolling widgets porting.
Patch by: Gustavo Lima Chaves <glima@profusion.mobi>
SVN revision: 73306
Subject: [E-devel] [Patch][elm_widget] access hook
Real time setting would be necessary in case of running applications
which
have elm_layout or elm_widget_item.
If the access mode does not care of this, running applications should
restart. The attached for this. Please review the patch and give any
feedbacks. Thanks.
SVN revision: 73229
Subject: [E-devel] [E-deve] [Patch] Add new apis for editable set/get
in multibuttonentry
Typically multibuttonentry is used as a composer.
But sometimes, Multibuttonentry can be used as a viewer.
Currently there is no way to use multibuttonentry as non-editable
viewer.
So I added new APIs which are elm_multibuttonentry_editable_set/get().
If editable is true, entry in multibuttonentry is activated and user can
compose item by pressing the return key.
If not, entry is hidden and muttibuttonentry works as a non-editable
viewer.
SVN revision: 73223
Subject: [E-devel] [Patch] Patch for normal scaling in elm_panel.c
widget
Problem:
Ticket #656 (new Bug) . The whole test window size will just grow
larger and
larger on each _theme_hook called.
Easy to reproduce: 1. Open elementary_test's panel test 2. Do
something that
issues a theme_hook (change scale etc)
SVN revision: 73076
Subject: [E-devel] [Patch][elementary] elm_access, use widget_item in
focus chain
It has been possible to use widget_item in access mode. The mouse-in
event
makes the widget_item speak.
But focus chain is not useful in access mode, because the widget_item
cannot get the focus as the edje object.
The accessibility highlight will move by keeping the attached patch
and you
can hear the sound of widget item information.
Please review the patch and give some feedbacks. There would be further
works such as dynamic setting focus next hook,
the previous _elm_access_list_register();, but please keep this first if
there is no big trouble. I'll do those also. Thanks.
SVN revision: 73074
Subject: [E-devel] Fwd: [Patch] elm_toolbar add signals patch.
For new style of toolbar (align icon and text to center of item) are
necessary some signals ("elm,state,text,visible",
"elm,state,text,hidden", "elm,state,icon,visible",
"elm,state,icon,hidden")
New style for toolbar, where icon and text are align to center of item.
SVN revision: 73028
One should either listen to mouse events and filter the corrosponding multi
events, or listen only to multi events. We chose the former, so we should
handle it correctly.
SVN revision: 72905
revert ->
Log:
From: Daniel Willmann <d.willmann@samsung.com>
Subject: Re: [E-devel] [PATCH] elementary: Add
elm_progressbar_maxunit_{set, get} functions
Some more discussion in IRC led to the conclusion that this should be a
handled in a signal callback I've attached the new patch where I also
updated the progress bar example.
Author: raster
Date: 2012-06-19 02:26:29 -0700 (Tue, 19 Jun 2012)
New Revision: 72459
SVN revision: 72506
Implements the module unloading to all modules on elm_module_shutdown.
This change also fixes a memory leak within elm_module picked by valgrind.
Here follows the valgrind log:
96 bytes in 1 blocks are definitely lost in loss record 105 of 168
at 0x4C29DB4: calloc (
in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
by 0x52065B6: _elm_module_add (elm_module.c:223)
by 0x520669E: _elm_module_parse (elm_module.c:77)
by 0x51F1249: elm_quicklaunch_sub_init (elm_main.c:437)
by 0x51F12BE: elm_init (elm_main.c:180)
Patch by Leandro Dorileo <dorileo@profusion.mobi>
SVN revision: 72499
Removes the event callback list within _smart_del. The current
implementation does not care about the node data resulting on
memory leaks.
Here follows the valgrind log:
16 bytes in 1 blocks are definitely lost in loss record 11 of 168
at 0x4C29DB4: calloc (in
/usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
by 0x522EC34: elm_widget_event_callback_add (elm_widget.c:1996)
by 0x5A7AA2B: _edje_emit_handle (edje_program.c:1322)
by 0x5A766B6: _edje_message_queue_process (edje_message_queue.c:764)
by 0x5A76856: _edje_job (edje_message_queue.c:160)
by 0x628C02A: _ecore_job_event_handler (ecore_job.c:115)
by 0x6288C1F: _ecore_event_call (ecore_private.h:319)
by 0x628D09B: _ecore_main_loop_iterate_internal (ecore_main.c:1814)
by 0x628D566: ecore_main_loop_begin (ecore_main.c:931)
by 0x4023B0: elm_main (test.c:198)
Patch by Leandro Dorileo <dorileo@profusion.mobi>
SVN revision: 72498
Subject: Re: [E-devel] [PATCH] elementary: Add
elm_progressbar_maxunit_{set, get} functions
Some more discussion in IRC led to the conclusion that this should be a
handled in a signal callback I've attached the new patch where I also
updated the progress bar example.
SVN revision: 72459
Thanks to Daniel Willmann pointing them out to me. Actually I wonder
why we need all this define and undef for ENGINE_COMPARE. Will dig a
bit deeper into that and see if we may can go with a single one.
SVN revision: 72144
strncmp() really does not like that. So we should not be lazy and check for it. Say thanks to
scan-build which reported these to me.
SVN revision: 72142
Subject: [PATCH] elementary: Minor fix to the documentation of ActionSlider
Date: Thu, 14 Jun 2012 15:31:05 +0100
Yeah, the most overhead I could imagine. :-)
SVN revision: 72141
Subject: [E-devel] [Patch] elm_config_font_overlay_apply bug fix
[Current Issue]
- The API applies the changes with elm_config_font_overlay_set().
But at the first time, application fonts have not been changed after
elm_config_font_overlay_apply api is called.
The application fonts changed only when the display is updated or
elm_config_font_overlay_apply api is called again.
[Main cause]
- When the elm_config_font_overlay_apply is called at fist time, it
calls
edje_text_class_del and edje_text_class_set in locally. But
edje_text_class_set just returns without _ejde_reclc() when new text
class
is ceated.
So in this case, there is nothing to be changed.
Since second call the api, edje_text_class_del api calls
_edje_recalc() in
locally. So the application fonts are changed.
[Change Description]
- There is two method. The first is to enable ejde_test_class_set api
is
can be recalced when the fist calling.
The second is to add _elm_rescale() function in
elm_config_font_overlay_apply API. This occurs edje_recalc effect
after font
overay set.
I attached the second method.
SVN revision: 72124
evas_map_point_coord APIs have the int values but evas_map_point_image_uv_set have the double parameters on the other hand.
This inconsistency causes the inexact result(flickering).
It's not good APIs for users.
SVN revision: 71979
ChunEon Park
PRINCE KUMAR DUBEY
Christopher Michael
Eduardo de Barros Lima
Carsten Haitzler
Gustavo Lima Chaves
Kim Shinwoo
Leandro Pereira
Jihyeon Seol
Vincent Torri
Gustavo Sverzut Barbieri
Daniel Juyung Seo
Cedric BAIL
Michael BOUCHAUD
Jonas M. Gastal
Jihoon Kim
WooHyun Jung
Jaehwan Kim
Seunggyun Kim
yan.wang
Tom Hacohen
Jiyoun Park
Minseok Kim
Artem Popov
Daniel Willmann
Bruno Dilly
Davide Andreoli
Stefan Schmidt
Mike Blumenkrantz