so edje was allocating 32 pointers per collection. this is per
collection inside an edje file even if we just use one collection from
that edje file. it consumes 32 pointers. on 64bit thats 256 bytes...
just for pointers to mempools so we can "optimize" freeing and
allocation of parts. this was simply rediculous. i moved it to a
sub-struct allocated on demand (so now only for collections we
actually use) and this nuked 400k of "base memory usage youcant get
rid of).
note that our current default theme has something like 1100 or so
images, 1500 or so collections in it. as theme gorws, memory footprint
goes up if we dont allocation only on demand (when needed/used) and we
aren't careful about the size of our data structs and their content.
@optimize
so ... Edje_Calc_Params was huge ... like about 200 bytes. every part
in every live edje object got one of these in addtion to real part
struct info etc. ... so really every part was probably consuming
300-500 bytes or so... crazy. so i made a lot of the data now optional
so only the minimum required is allocated now which cuts down about 110
or even 120 bytes per part, depending. 100 bytes was needed for 3d
node parts even though almsot no parts are 3d node parts... the image
and text data was 30-40 bytes so we consumed 100 even if we only used
30-40... so this cuts that done and puts in polace calc param cleanup
funcs everywhere they are needed to clean up this extra allocated data.
i also reduced this even more by maping pointers to req_drag, map and
physics and clip_to fields in another extension struct cutting
down another 28/52 bytes on most parts (in return for an added
4/8 bytes - on 32/64bit accordingly).
in elementary_test this saves about ~300kb of ram for just having the
etst run and displaying (peak memory measuremment). so massif says
10.6M -> 10.3M.
@optimize
so we didnt set everything to null after being freed and object sae
freed after some data. do the data frees after objects are deleted so
callbacks cant access null data anyway. this makes this edje shutdown
far more robust and safe. massive improvement in stability i hope.
this saves about another 80Kb or so in dirty pages by only loading
ephysics when needed. This removed ephysics and bullet library dirty
pages from the process space. this is another patch to address T4227.
@fix
We now use eo_do() and EFL_CANVAS_SCENE3D_CLASS macro to create viewport
Reviewers: raster, Hermet, cedric
Subscribers: jpeg, artem.popov
Differential Revision: https://phab.enlightenment.org/D4163
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
Coverity reports that 'text' here is a null pointer dereference so
check for valid 'text' variable before trying to use it.
Fixes Coverity CID1267490
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: check if the size of scene is bigger than 0x0 and build 3D scene in the edje_player in this case and use "opengl_x11"
Reviewers: Hermet, jpeg, cedric
Subscribers: NikaWhite, Oleksander, artem.popov, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4041
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
this should cut some memory used by edje by using smaller types like
shorts instead of ints where we just dont need a full int range and
short will do, and re-ordering in memory data soit packs better when
accoutning for alignment
Summary: Creation of scene and root node in edje-3d with all 3D-parts of edje object. Add some new methods to edje_util.c
Reviewers: raster, Hermet, jpeg, cedric
Reviewed By: cedric
Subscribers: artem.popov
Differential Revision: https://phab.enlightenment.org/D3963
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
This reverts commit 546ff7bbba.
It seems that eo_del() is useful and removing it was creating bugs.
The issue is that the way we defined parents in eo, both the parent and
the programmer share a reference to the object. When we eo_unref() that
reference as the programmer, eo has no way to know it's this specific
reference we are freeing, and not a general one, so in some
circumstances, for example:
eo_ref(child);
eo_unref(child); // trying to delete here
eo_unref(container); // container is deleted here
eo_unref(child); // child already has 0 refs before this point.
We would have an issue with references and objects being freed too soon
and in general, issue with the references.
Having eo_del() solves that, because this one explicitly unparents if
there is a parent, meaning the reference ownership is explicitly taken
by the programmer.
eo_del() is essentially a convenience function around "check if has
parent, and if so unparent, otherwise, unref". Which should be used when
you want to delete an object although it has a parent, and is equivalent
to eo_unref() when it doesn't have one.
This should now fix the part API usage once and for all.
EFL should have no part name in any of its APIs beyond
the Efl.Part interface.
Part proxy objects (may be real objects) have a lifetime
of only one function call, in a fashion similar to eo_super.
@feature
We used to have eo_del() as the mirrored action to eo_add(). No longer,
now you just always eo_unref() to delete an object. This change makes it
so the reference of the parent is shared with the reference the
programmer has. So eo_parent_set(obj, NULL) can free an object, and so
does eo_unref() (even if there is a parent).
This means Eo no longer complains if you have a parent during deletion.
Summary:
According to D3710 new field "camera" in edc was added for IMAGE parts.
It is the name of the CAMERA part to set its viewport as a source of image if no image name is given.
Reviewers: raster, Hermet, cedric
Reviewed By: cedric
Subscribers: jpeg, artem.popov
Differential Revision: https://phab.enlightenment.org/D3777
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
This fixes on top of 49a27688b1, which
assumed somehow that table items had names, although that
might not be the case. In my situation, name = NULL and there
was a crash everytime I clicked on the clock widget.
If the table item has a name, posible case when item name length, with
index, will be 12. The 12 is predefined length for box index.
Quote Cedric
In a box, the index is one dimension, one int, thus the length
of it (from int to string) will always fit inside 12 bytes. That's
where this 12 comes from. That's also how the unique name of that item
is defined.
This commit separate the items name generation by part type, it will be
more correctly.
@fix
It has been decided that we would not use any namespace for interface
and they will sit in efl main namespace.
This patch doesn't correct the naming of the event has we don't have a
prefix for event. We do still have EFL_ANIMATOR_EVENT_ANIMATOR_TICK,
instead of a nicer EFL_EVENT_ANIMATOR_TICK.
I just ran my script (email to follow) to migrate all of the EFL
automatically. This commit is *only* the automatic conversion, so it can
be easily reverted and re-run.
Summary: Initiation of Evas Objects and creation of nodes for new part types in Edje.
Reviewers: cedric, raster, Hermet
Subscribers: jpeg, artem.popov
Differential Revision: https://phab.enlightenment.org/D3665
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
The "color_tree" block contains a list of one or more "node" blocks.
A "node" block consists of its own color class name and the list of child
color classes. At runtime, parent color class will be referred instead,
if child color class is set to part but its color values are not defined.
Reviewers: raster, Jaehyun_Cho, jpeg, cedric
Reviewed By: cedric
Subscribers: cedric, kimcinoo
Differential Revision: https://phab.enlightenment.org/D3606
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
This is necessary for backward compatibility still I am thinking of displaying a warning for this
use case and request people to update there work and drop that feature in the future (In a year
maybe from now). Elementary doesn't need this as it depends on Ecore_Evas.
Edje_Part can change its min or max size in code level with
size_class.
Differential Revision: https://phab.enlightenment.org/D3329
PS: Manual commit, arc refused to work...
@feature
Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
evas_object_clipees_has is far cheaper than evas_object_clipees_get in case of checking if
clipees exist or not. This should improve the performance in case of large set of clipees.
@fix
Without that, the image has no fill information. Fill properties
may need to be added to SNAPHOT parts but the default behaviour
should make sense. Before this patch you just get a black rectangle.
Considering how image filters currently work, marking snapshots
as filled by default is not the best solution (need padding_set(0)
to render nicely).
This is now like the other embedded scripts, where a verbatim
string is parsed. The syntax is now:
filters {
filter {
name: "filter0";
file: "filter.lua";
}
filter {
name: "filter1";
script {
blend {}
}
}
filter.file: "file.lua"; // name is "file.lua"
}
Thanks @raster for the quick review.
Deep down internally there was already a name, but no API could
really set it properly.
Here Edje will set the name of the filter based on the part name
or the data item name if relevant.
Use the file data {item, file} block to embed filters code.
It can become especially useful to keep the filters as separated
Lua files, that will be embedded in the final edj file.
@feature
Those objects should never be rendered on the canvas, even if they
are visible. On the other hand, they need to be rendered in mask or
proxy surfaces.
note: this patch includes some extra whitespaces changes :(
@feature
Since masking, for performance and themeing reasons, it becomes
interesting to be able to switch clippers on the fly. In particular,
switching from an IMAGE mask to a standard RECT clipper can save a
lot of power when masking is not required.
This new flag "description.clip_to" will behave a bit like a mix of
rel.to and visible:
- It points to a part by name, just like part.clip_to. This will
override the clipper set by the part, or override the default clipper.
- Like "visible", it can not be interpolated between two values, so
it will switch only at the end of a transition.
- By default there is no clip override, which means Edje will fallback
to the part's clipper, if any, or the base (group's) clipper.
NOTE:
- Since a clipper that does not clip anything becomes a standard object,
it is visible and rendered. This will in 99.999% cases not be the
intended behaviour. Currently we can simply use a transparent RECT
in order to always have something clipped by the clipper, but this is
a hack and this will trigger rendering of masks in their surfaces even
when they are not actually used.
Ideally, there should be a flag indicating to Edje & Evas that an object
should be considered a clipper in all situations, and never be rendered
on screen.
TODO:
- Support Edje Edit
- Add Embryo & Lua functions
- Add support in edje_convert
- Add Edje/Evas flag to mark objects as "no_render"
@feature
Summary:
Implementation to support .po files in edc for translation
Test Plan:
Test Code to test this implementation is done as part of efl/src/examples/edje/edje-text.c and efl/src/examples/edje/text.edc
edje_cc -md . text.edc && gcc -o edje-text edje-text.c `pkg-config --libs --cflags ecore-evas edje evas ecore eo`
./edje-text
1) Click On the text "Click here"
The language gets changed.
Reviewers: shilpasingh, cedric
Reviewed By: shilpasingh, cedric
Subscribers: cedric, rajeshps, govi, poornima.srinivasan
Differential Revision: https://phab.enlightenment.org/D2573
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
This reverts commit c38f84e64f.
apparently many existing edje groups were written with the assumption that
this was broken, so fixing it is impossible at this point
previously these parts would fail to consume mouse events as expected,
leading to strange event chains which were inconsistent with other types
of parts
@fix
Summary:
Internationalisation of the static text specified as part of the edc is implemented.
Problem: Static text when specified in the edc, remains unchanged when the system language is changed.
Solution: Language support is provided even for the static strings in the edc.
Test Plan:
Test code to test this implementation is done as part of efl/src/examples/edje/edje-text.c and efl/src/examples/edje/text.edc
Compile the code with the below command
edje_cc -md <dir path>/efl/src/examples/edje/ text.edc && gcc -o edje-text edje-text.c `pkg-config --libs --cflags ecore-evas edje evas ecore`
./edje-text
1) change the language of the system using the command
export LANGUAGE=hi
./edje.text
Not the text Loading gets displayed in hindi language
2) change the language of the system using the command
export LANGUAGE=ta
./edje.text
Not the text Loading gets displayed in tamil language
3) change the language of the system using the command
export LANGUAGE=en
./edje.text
Not the text Loading gets displayed in english language
As the number of .mo files in the /edje folder can be increased, those many languages can be supported
Reviewers: cedric, shilpasingh
Reviewed By: shilpasingh
Subscribers: cedric, rajeshps, govi, poornima.srinivasan
Differential Revision: https://phab.enlightenment.org/D2336
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
When textblock styles have text_classes, all edjes in the files were added
to text_class_member_hash even if the edjes didn't use the textblock styles. It
makes time long to update text_class.
This will add the edje using the textblock style which has a text_class to
text_class_member_hash.
Reviewers: cedric, raster
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D2035
thjis was deprecated even before efl 1.0 by never removed. lua
replaced it for script_only objects and you havent been able to
compile an edje file with script_only enabled since 1.0, so no point
having the code here.
this cleans up that code and cruft.
so this has to go. reloading edje files is nothing but trouble.
example - if you update your os your theme files get updated and then
all sorts of stuff goes wrong. jeff is right. this makes it an
intractible problem. we have an open file handle on the edj file. we
share that anre reuse it via eina_file - keep it. this keeps tyhe edje
file stable and consistant.
<Jef91> Elementary applications freak the fuck out
<Jef91> if you change the theme file
<Jef91> that they are using
<Jef91> SeoZ,
http://forums.bodhilinux.com/index.php?/topic/10629-eepdater-display-issue/
<Jef91> that happens when my theme file that is in use changes
we will get nothing but continued issues and complains if we keep
doing this. it's a fairly pointless mis-feature. thank god its got no
apis - just signals and internals.
this massively improves edje performance when using groups, which previously would continue calculating their parts even when their parent object was hidden
CPU usage in my test case went from 20-30% to 1%.
@fix
Summary:
...cannot encode those things into edje.
In our case, we need vibration when longpressed. But those files are not
audio or image and cannot be encoded into edje. Also, this library is not
opensource so should not be linked directly with edje.
So we should call vibration API by using this plug-in.
Reviewers: raster, cedric, seoz, Hermet
CC: cedric
Differential Revision: https://phab.enlightenment.org/D588
Add to fuction prototype new param: Eina_Bool approximation.
If need exact matching state name and value set EINA_FALSE to
'approximate'. In other cases used EINA_TRUE.
Reviewers: cedric, raster, seoz
CC: cedric
Differential Revision: https://phab.enlightenment.org/D400
Signed-off-by: Cedric BAIL <cedric.bail@samsung.com>
With Eina_File we now can pass an efficient handler accross library boundary. Edje
and all underlayer already use it to avoid race condition when setting an Edje object.
Elementary and Enlightenment are still exposed to some potential race condition when
an Edje file is modified underneath there feet. With the following set of function it
should now be possible to avoid those race condition to:
edje_mmap_data_get
edje_mmap_collection_list
edje_mmap_collection_list_free
edje_mmap_group_exists
Carsten Haitzler
Jee-Yong Um
Tom Hacohen
Subhransu Mohanty
Stefan Schmidt
perepelits.m
Chris Michael
Jean-Philippe Andre
Jean Guyomarc'h
Jean-Philippe ANDRE
Cedric Bail
Vyacheslav Reutskiy
Mike Blumenkrantz
Jee-Yong Um
Amitesh Singh
katpavalli
Andrii Kroitor
Sohyun Kim
ChunEon Park
Mike Blumenkrantz
zmike
Yossi Kantor
Tae-Hwan Kim
Albin Tonnerre
Cedric Bail
Cedric Bail
