It has been discussed on the ML (thread: "[RFC] rename efl_self") and
IRC, and has been decided we should rename it to this in order to avoid
confusion with the already established meaning of self which is very
similar to what we were using it for, but didn't have complete overlap.
Kudos to Marcel Hollerbach for initiating the discussion and
fighting for it until he convinced a significant mass. :)
This commit breaks API, and depending on compiler potentially ABI.
@feature
As far as I remember, declaring structures and arrays in a cast is a GCC
extension. I'm not 100% sure I'm right, but I remember it was the case.
Regardless of whether it's an extension or not, this commit removes that
pattern and makes everything cleaner (and faster?).
This is another follow up to the investigations of T4227. As stated
there, in any PIE (a shared library is one), structures, even const ones
end up being written to because of dynamic relocation. This means that
using static const structures has actually lead to no savings, only
waste. Since we never really needed them, using them made things even
worse than just having a different API that doesn't save them.
Thus, this commit changes the way we set the functions. Instead of
passing a pre-populated struct, we now just have an initialiser function
where you set the functions. This on its own doesn't significantly reduce
the amount of dirty memory pages for a reason I have yet to uncover,
though I believe it's done as a misguided compiler optimisation.
However, this design is flexible enough so we can change to another one
that is quite ugly, but I have already tested and proven that does that.
This patch series doesn't include the better improvement (passing
everything on the stack as va_args) because the API was too ugly
for me to bear, and I would rather first make sure there is no way to
force the compiler to do the right thing here.
Unfortunately this commit gives up on useless stricter validation.
Before this commit we would make sure that we are only overriding
functions correctly defined in our hierarchy. With this one, we don't
anymore. This is not a big problem though because this is a check that
is also enforced by Eolian. So as long as you are using Eolian, you
should be fine.
Breaks API and ABI!
@feature
Summary:
There can be the case that the file of a edje is NULL. Even if this case
is a bit strange, we should not crash on it.
Sample code which produces the crash:
Edje_Object *edje;
int r, g, b, a;
edje = edje_object_add(evas);
edje_obj_color_class_get(edje, "bla",
EDJE_COLOR_CLASS_MODE_COLOR, &r, &g, &b, &a);
So better protect against this case.
Reviewers: raster, herdsman
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4277
This improve speed of processing events in genlist scrolling benchmark by 30%
inside the efl_object_event_callback_call code. Not a really big deal as it
goes from 0.9% to 0.6% of the total time spend. Welcome to micro optimization.
The Efl.Net.Dialer.Websocket is just like other Efl.Net.Dialers: you
can dial, you can close, monitor connected/address resolved and so
on. And you can use WebSocket primitives and events such as
text_send(), binary_send(), ping() and close_request() (since
WebSockets use a close process where you should state a close
reason). See efl_net_dialer_websocket_example.c
Even if WebSocket is a message-based protocol (like "packets" from
UDP), you can use efl_net_dialer_websocket_streaming_mode_set() to
tell it to handle text or binary messages as a stream. Then all the
Efl.Io.Reader and Efl.Io.Writer APIs work as expected, see
efl_io_copier_example.c updates.
When CURLOPT_WRITEFUNCTION returns less then the requested amount,
CURL will fail, not call us back with the remaining data.
Then in such cases we must pause CURL and read nothing.
When unpausing we need to kick curl with timeout action so FD handlers
will be re-arranged.
Last but not least, sync our buffer limit with CURL, otherwise it may
always fail if we're smaller than CURL.
CURL doesn't play nice if handles are deleted or modified while it's
dispatching the callbacks, then we must not touch the CURL* easy
handle in those cases, just dissociate the handle from object and
schedule a job to do the deletion later.
Also, since from CURL callbacks we do not have the reference to the
object, if they are deleted from inside the callback, users of 'pd'
will crash. Thus keep an extra reference while the object and its
private data are in use.
The curl_multi_info_read() is used to notify of errors and
end-of-stream, if we do callback directly from there, the user may
efl_del(dialer), which will result in the "pd->easy" being destroyed
with curl_easy_cleanup() then "cm" and "cm->multi" being destroyed.
Thus postpone that action and keep a list of finished objects, calling
their event handlers which can delete the object (or siblings), thus
ref before dispatching and unref afterwards, taking care to monitor
EFL_EVENT_DEL so we do not use stale objects.
while in a job we do not have the safety of eo holding us alive and
when we call back the user, he may have deleted the object, releasing
both the object and its private data that we're using.
then keep an extra reference, call the methods and release it.
This widget was lacking an edje_external param to notify the widget to start or
stop pulsing from edje.
@fix Now the edje_external progressbar test works as expected.
Summary:
The port reuse feature and 'SO_REUSEPORT' flag are not supported by a few linux
In case of linux kernel, it supported from v 3.9
(https://kernelnewbies.org/Linux_3.9)
On the lower version of kernel, compile is failed
Reviewers: barbieri, jayji
Reviewed By: jayji
Subscribers: akanad, id213sin, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4256
Summary:
Previously, focused item was always selected on first and last key action.
If item_select_on_focus_disable is true, item should not be selected.
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
Signed-off-by: jinwoo.shin <jw0227.shin@samsung.com>
Test Plan: elementary_test -to genlist
Reviewers: cedric, SanghyeonLee, Hermet
Reviewed By: Hermet
Subscribers: sju27, seoz, jpeg
Differential Revision: https://phab.enlightenment.org/D4272
Summary:
"/**" requires for doxygen, but one "*" is omitted for the reference
of eina_matrix3_multiply().
Reviewers: Hermet
Reviewed By: Hermet
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4266
This combines evas canvas functions to list and query
touch points into a single iterator:
- evas_touch_point_list_count
- evas_touch_point_list_nth_xy_get
- evas_touch_point_list_nth_id_get
- evas_touch_point_list_nth_state_get
This also fixes a number of issues related to feeding fake
input events.
Note: I wanted to add delta x,y information as well but it's
in fact not really possible outside the event callback itself,
as the previous x,y position will not be updated unless there's
an event.
@feature
Those two properties aren't related to a "drawing" canvas
but to the current state of input.
Note: both Efl.Input.Pointer (pointer input event data) and
Efl.Input.Interface (common interface for input handling objects)
expose a pointer position API. Not sure what to do about that.
This adds support for distance, pressure, tilt and twist.
Not entirely sure if normalized & raw (x,y) should be exposed
in the eo interface. Also not sure what to do with tilt_x/y
(as used by libinput) or touch/tool width "major/minor" vs.
radius x/y.
Add debug logs in the example, including the distance.
I can't test most of these values due to a lack of compatible
hardware, but the most basic features seem to work :)
First, fixing ellipsis text positions: ellipsis items should be assigned the
text positions of the omitted text (while maintaining the formatting of the
last visual item). In the case where an entire item was rejected, it
will be assigned that item's text position. If an item was split, it will be
assigned the text position of the split portion.
The BiDi reorder code relies on properly-assigned text positions.
Second, fixing ellipsis handling: the width calc was only considering the
ellipsis item's width. However, if the ellipsis is placed as e.g. the first
visual item (such as in RTL cases), its advance value should've be considered,
instead.
Thanks Youngbok Shin for the test case and information.
@fix
Summary:
These macros allow you to define module informations like
author/description/version/license
e.g.
// Use "Name <email id>" or just "Name"
EINA_MODULE_AUTHOR("Enlightenment Community");
// Mention license
EINA_MODULE_LICENSE("GPL v2");
// What your module does
EINA_MODULE_DESCRIPTION("This is what this module does");
// Module version
EINA_MODULE_VERSION("0.1");
Now eina_modinfo can show these informations to users
$ eina_modinfo module.so
version: 0.1
description: Entry test
license: GPLv2
author: Enlightenment Community
@feature
Reviewers: cedric, tasn, raster, jpeg
Subscribers: seoz
Differential Revision: https://phab.enlightenment.org/D4257
After my many input events changes, a same object callback
could be called multiple times in a row because both mouse
and multi events were sent. As such, the multi event had no
direct effect (no callback called) but it reset the object's
last event type. This allowed the mouse event callbacks to be
called again.
Note that I haven't tested multi touch yet :(
Very good catch by @bu5hm4n!
Fixes T4462
Fixes T4467
enventor exhibits this issue, where the focus target is
NULL in some cases. The ERR message was harmless, but it's
good to avoid it and be explicit that the object should be
non NULL when adding event callbacks.
Since pointer events are all the same thing, users may not know
what values are valid for what kind of event. Eventually we
want to expose more information, but we also need a way to inform
the caller about the validity of the values we get.
I guess (can't test) that multi touch was broken, as the
position of the event was set to the position of the pointer
on the canvas. Which means all fingers would be in the same
spot, no matter what the real input. Copy & paste error.
provide curl with CURLOPT_OPENSOCKETFUNCTION and keep the fd in our
private data.
This is required because on _efl_net_dialer_http_efl_io_writer_write()
we may have no fdhandler.
It happened to me while implementing the WebSocket that uses a
bi-directional communication on top of HTTP and the server sent the
whole message, CURL reads:
recvfrom(7, "...", 16384, 0, NULL, NULL) = 86
recvfrom(7, "", 16384, 0, NULL, NULL) = 0
After the empty (second) recvfrom(), CURL will remove the fdhandler:
DBG:ecore_con lib/ecore_con/efl_net_dialer_http.c:482 _efl_net_dialer_http_curlm_socket_manage() dialer=0x4000000040000005 fdhandler=(nil), fd=7, curl_easy=0x5561846ca8d0, flags=0x4
However I should be able to write to this socket, in my case I need to
reply to a PING request with a PONG.
CURL is smart and when you ask for CURLOPT_HTTPGET, it will
automatically configure UPLOAD=false. Likewise, if you ask for
UPLOAD=1 it will configure CURLOPT_PUT...
However, to do things like WebSocket we need to do a GET request where
we need to send data, then UPLOAD=true must be used.
Then use both information in order to setup the request method and
upload, using CURLOPT_CUSTOMREQUEST to force a given HTTP method.
If we delete the curl multi handle, then we should stop any timer that
was scheduled, otherwise it will use a dead or null pointer.
also add some debug to help track down when the multi handle is
deleted.
emile_binbuf_sha1() was actually doing HMAC version using the given
key. This doesn't work when all you need is just the SHA1 of the input
data.
Then rename emile_binbuf_sha1() to emile_binbuf_hmac_sha1() and
introduce a new version without key/keylen.
This API was marked as BETA and no real users in the codebase, then it
shouldn't cause us problems.
eina_error_msg_get() must return NULL if an incorrect error is provided.
The XSI strerror_r() returns EINVAL when an invalid error is passed to
it, so we can end the function here. If we kept on, we would have tested
against the 'unknown_prefix' ("Unknown error ") which is implementation
defined, and registered a new error when the invalid error message
didn't match the 'unknown_prefix'. This new error message would have
been returned, which is not what we expected.
This case arised on Mac OS X where the 'unkwown prefix' is
"Unknown error: " instead of "Unknown error ".
It fixes eina test suite on Mac OS X.
Use the new behavior of Efl.Object.event_callback_call to correctly
update events to pass Efl.Model objects while still suppling path
strings for legacy smart callbacks.
Override Elm.Fileselector.event_callback_legacy_call in order to separate
the types of any incoming event call that uses Efl.Model.
Summary:
When selection is there we are passing the whole text and position to imf
which sees that next character like matras (eg .Hindi) should be inserted
(if pressed) and when matra comes the selected text is gone and only matra
remains in the entry.
eg: we have text in hindi like मानक, select all and hit matra in keyboard
selected text is gone, only matra is there.
@fix
Test Plan: Tested in Tizen device
Reviewers: jihoon, tasn, herdsman, thiepha
Subscribers: Hermet, shilpasingh, raster, subodh, jpeg, cedric
Differential Revision: https://phab.enlightenment.org/D2951
also help reduce fragmentation. also remove callbacks immediately if
callbacks are not being walked at the time (as opposed to just marking
them to need deletion then call a clean that if not being walked will
walk all cb's when we already know what to remove).
@optimize
Focus highlight can be on the air, when focused object in scroller is
moved by mouse down and move.
So, mouse up is not proper for executing focus_auto_hide.
@fix
The visible bug for this issue is that the Elm External Video example
is broken:
elementary_test -to "ExtVideo"
The root cause is that the "file,choose" event of the
FileselectorEntry is (wrongly) not listed by the function:
evas_object_smart_callbacks_descriptions_get()
This evas functions is used in elm_external to forward all the widget
events to edje, but it cannot forward the "file,chosen" event because
it's not listed.
Thus the video test is not working for the lacks of that event.
I think the fix should be somewhere in elc_fileselector_entry.c, there
are some hacks there for the incriminated signal, but I don't know how
to properly fix.
as pointed out by dave:
DaveMDS added a comment.Fri, Aug 26, 5:19 PM
I think the problem is in this function: (elc_fileselector_entry.c)
...
this fixes T4337
In a case where eina_promise_then is executed immediately (like with some
quick and light Efl.Model), the Listing_Request struct will be prematurely
freed in the first iteration of the child processing loop, because the
item_total counter had not accumulated the right number of items yet.
With this commit, we traverse the children accessor first, so we can know
the number of items.
Also, no longer use the Listing_Request pointer after the loop, once it
may have been deallocate already.
And put a note about this too.
This one is a bit tricky... When we create the aggregated
promise, if one of the properties of the model returns an
error, the eina_promise_then() will immediately call the
error callback. In this happened for the first item, the
total items in the listing request would be 1.
Before this commit, we tested for incremented the processed
counter and compared it to this total count. If it was
greater or equal, we would free the common listing request.
But in the case of successive failures, we would set the
total counter to 1, then the processed counter to 1 and
therefore free. Then increment the total counter to 2, then
then processed counter to 2, and free again... which would
cause an abort() from the libc or something else nasty.
Now we just decrease the total count of items. We avoid
the cases and double frees, without leaking.
The carray iterator will end iterating only when it finds a NULL
object. We must make sure the last element of the array is NULL
to avoid out of bounds access.
Mac OS X does not support POSIX unnamed semaphores, only named
semaphores, which are persistant IPC: when the program exits,
and if semaphores where not released, they stay forever...
All EFL programs were "leaking" a semaphore, due to how
eina_log_monitor manages its resources. Therefore, after building
EFL a lot (which run eolian_gen, eolian_cxx, elua, edje_cc, ...)
we were not able to create any semaphore...
Now, we get rid of these semaphores and use Mac OS X's own
semaphores. Code is less cumbersome, and we don't have any
disavantage of the named semaphores.
Fixes T4423
@fix
Efl.Object.event_callback_call no longer calls legacy smart callbacks;
calling only event callbacks registered with the given event description
pointer.
Create the method Efl.Object.event_callback_legacy_call to inherit the old
behavior from Efl.Object.event_callback_call, calling both Efl.Object events
and legacy smart callbacks.
Update all other files accordingly in order to still supply legacy
callbacks while they are necessary.
This simply avoids calling functions on NULL objects, since
the previous patch would ERR out rather than silently ignore
the problem.
I just add explicit NULL checks before calling the functions,
so it's clear the object could be NULL (in the widget).
This removes useless magic checks (only check whether the
arg is not null) that were not even present in every function.
The cost should be similar or lower than an eo function call.
This removes:
Efl.Event interface
And renames:
Efl.Event.Input -> Efl.Input.Event
Efl.Event -> Efl.Input.Event (merged)
Efl.Event.Pointer -> Efl.Input.Pointer
Efl.Event.Key -> Efl.Input.Key
Efl.Event.Hold -> Efl.Input.Hold
This also moves some interfaces from efl/ to evas/ where they
belong better.
This allows renaming Eo_Event to Efl_Event.
This is for Wacom graphics tablets (with a pen).
The raw data sent by ecore to evas (and then to apps) is pretty
useless as it's not normalized, and apps have no way of knowing the
dimensions of the tablet, without themselves opening the device
(we don't know nor expose the path to the device).
This is for Xi2 only for now, as Wayland support hasn't been done
yet.
The intent is to deprecate LABEL_X and LABEL_Y. I'm not sure yet
if the normalized value is useful or not (it would seem we may not
be able to provide this info in Wayland).
The new WINDOW_X, WINDOW_Y labels will be used in the new event
type (Efl.Event.Pointer). Normalized values are not exposed yet,
let's decide if we want them or not first (based on what can be
done in Wayland space).
@feature
Mice in X with xi2 send Axis events which are badly defined,
and carry basically useless information, as we also receive
proper mouse events. Notably, all mice input events are
"Rel something" but in fact they are absolute values (even
the wheel information is a counter increasing every time you
scroll).
This should not break any application as such axis events
carried only values with label ECORE_AXIS_LABEL_UNKNOWN.
This also fixes a leak when n == 0 (no "valuator" found
in the list, this used to be unlikely, now happens at every
mouse event).
This converts Evas_Axis or Ecore_Axis info arrays into basic
pointer data. Also marks those fields as set. All events need
to properly implement the value_has property (mark all bits
whenever a value is known).
This sets a bit whenever a callback listener is added.
I couldn't get any profiling data easily (too small for
valgrind).
Note: This removes the proper refcounting on the "move"
event listeners. I believe this is not a problem as most times
the move_ref goes to 0, it is because the object is deleted.
Worst case, we just trigger a callback_call with no listeners.
This adds 32 bits to each evas object private data.
This event should not be exposed at all, it's not necessary
anymore, EFL_EVENT_DEL already exists and should be good enough.
This does move the callback call a little bit ealier in the del
process, but at first glance, this shouldn't have any impact.
This moves MULTI events to those new finger event types,
and also sends a finger event for finger 0 (aka the pointer).
NOTE: This may require a separation between a mouse input and
an actual finger touch. To be defined, ie: do we let the app
check the input device info to decide whether the event is
actually the first finger of a multi touch device, or do
we want to send only actual finger events from multi touch
devices only?
@feature
This is getting trickier, as those events have a lot more
side effects and complexity than a simple wheel event...
Some code has been added that should be fixed in the following
commits.
So glibc has decided that readdir_r is hard to use safely and deprecated it
this summer. They recommand to use readdir, which was in the past unsafe to
use in a multi thread scenario, but is now on most system (and all system
we care, including our own implementation in evil). It is basically safe
as long the same DIRP is not accessed from another thread. This is true in
our code base, so we are fine to go with this.
For further reading: https://lwn.net/Articles/696474/
It's entirely possible that a system doesn't have a cpu 0, so
when we try to pin all our render threads onto processor 0 we
may fail.
This results in some very connfusing build breakage when
edje_cc hangs up because its render thread didn't start.
So, if starting the thread with affinity fails, let's try without
affinity.
(This is trivial to reproduce - just use sysfs to turn off cpu0
after boot.)
@fix
This fixes argb windows transparency in E software compositor.
My current problem is that I have no idea what changed, why this
is needed now, and how things could actually work before.
Fixes T4389
@fix
so ok - efreet crashed letting you know you have a missing mimedb...
return NULL instead fixes T4425 :) - rememebr to kill your efreetd's
to get a new mimedb - log out and in will do that.
The use of low-level interfaces such as Efl.Io.Reader and
Efl.Io.Writer are not that user-friendly as they can handle partial
data.
Classes such as Efl.Io.Copier makes them easy to use, but they need a
reader (source) or writer (destination) and in our examples we used
fixed buffers or some existing streams (stdin/stdout/stderr,
networking...).
However, if interactively we need to produce some data to be sent,
such as implementing some networking protocols, we'd have to write our
own Efl.Io.Reader and Efl.Io.Writer classes to handle the buffering.
Not anymore! With Efl.Io.Queue you can write stuff to it and it will
buffer to memory. Once stuff is read, it will automatically remove
those bytes from buffer.
So actually there is quite a big issue with semaphores
on OSX. We use (named) POSIX semaphores, but this was
a (my) mistake... I'll fix it later...
The real issue is that named semaphore are persistants:
when the program dies, it stays alive. This is pretty
bad with eina_debug_monitor because we create a semaphore
we never release, due to a wild thread...
This leak of semaphores went unnoticed before commit
4a40ff95de because the
name of the semaphore was unique per process, and
overriden when another process was launched. This
was very bad, but saved us from overflowing the
semaphore pool. It is now overflowed pretty fast when
building a lot EFL, because of Eolian that runs A LOT!
So that's one problem that still needs to be fixed,
by using OSX' own semaphores (see T4423).
Another big issue, which is now fixed is that the
buffer in which we generated the semaphore ID was
too small, and therefore we were reduced to one shared
semaphore for a whole process... This buffer has been
now set to 31 characters, which seems to be the maximum
length of a semaphore ID.
So now things are better, but still with a deadly issue.
TCP_CORK is Linux only. TCP_NOPUSH is supposed to
do the same thing than TCP_CORK, but on BSD (including
Mac OS X).
We now check for the existance of TCP_CORK or TCP_NOPUSH,
and use the right option. If none exist, cork_{set,get}
will just fail.
This was never used and there is no plan to ever use it. I'm going to
soon add a different mechanism with which it will be possible to provide
them again to Eo if ever needed without breaking ABI. Though it's
unlikely it will ever be.
As pointed out in the mailing list, it was introduced in this release,
so it's better to remove the symbol instead of deprecating it.
People should use ETIMEDOUT directly.
so efreet mime was loading a bunch of mime type info files, parsing
them on startup and allocating memory to store all this mime info -
globs, mimetype strings and more. all a big waste of memory as its
allocated on the heap per process where its the SAME data files loaded
every time.
so make an efreet mime cache file and a tool to create it from mime
files. mmap this file with all the hashes/strings in it so all that
data is mmaped once in memory and shared between all processes and it
is only paged in on demand - as actually read/needed so if your
process doesnt need to know about mime stuff.. it wont touch it anyway.
this saves about 240-300k or so of memory in my tests. this has not
covered the mime MAGIC files which still consume memory and are on the
heap. this is more complex so it will take more time to come up with a
nice file format for the data that is nicely mmaped etc.
@optimize
This class implements the Efl.Net.Dialer interface using libcurl to
perform HTTP requests. That means it's an Efl.Net.Dialer,
Efl.Net.Socket, Efl.Io.Reader, Efl.Io.Writer and Efl.Io.Closer, thus
being usable with Efl.Io.Copier as demonstrated in the
efl_io_copier_example.c
Efl.Net.Server defines how to accept new connections, doing the
bind(), listen() and accept() for protocols such as TCP.
Efl.Net.Dialer defines to to reach a server.
Both are based on Efl.Net.Socket as communication interface that is
based on Efl.Io.Reader, Efl.Io.Writer and Efl.Io.Closer, thus being
usable with code such as Efl.Io.Copier.
The Server will emit an event "client,add" with the established
Socket, which is a child and can be closed by both the server or the
user.
The Dialer extends the Socket and allows for creating one given an
address, that will be resolved and connected.
TCP is the initial implementation so we an validate the
interfaces. UDP, Unix-Local and SSL will come later as derivate
classes.
The examples are documented and should cover the basic principles:
- efl_io_copier_example can accept "tcp://IP:PORT" and will work as a
"netcat", can send data from socket, file or stdin to a socket,
file, stdout or stderr.
- efl_net_server_example listens for connections and can either reply
"Hello World!" and take some data or work as an echo-server,
looping back all received data to the user.
More complex interactions that require a "chat" between client and
server will be covered with new classes later, such as a queue that
empties itself once data is read.
These interfaces allows generic operations on objects that can store
or provide data, such as a file or a buffer.
With well defined interfaces and events we can create code such as
Efl.Io.Copier, that will link a source with a destination and
progressively copy data as they appear.
Summary:
Copying from string 'buf' of length 4095 to '&socket_unix.sun_path[0]'
may form a non-terminated C string of size 108. So added null termination.
Signed-off-by: Prateek Thakur <prateek.th@samsung.com>
Reviewers: cedric, thiepha
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D4247
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
So, first, the wrong strerror_r() was detected on
Mac OS X. Instead of using a complex set of macros
to try to detect which strerror_r() to use, when
it is defined, let the autotools handle that clerverness
for us.
so every collection we add to the eina has of collections... we just
did eina_hash_add - this isn't quite good. because we use eina_file
and mmap the edje file. we COULd use the strings straigh fromt he edje
filer dictionary and not actually allocate private heap memory - thus
sharing all those strings. this was silly, and this saves about
another 120k of memory with the default theme as it has about 1500 or
so collections in it... and these strings add up fast.
@optimize
it seems vector type parts were not handled all that well. we had at
least one mem leak with the vector mempool never being freed... so i
filled in various missing points where vector parts were not being
handled right.
@fix
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
this speeds up downscaling of images by somewhere between 1.8 to 3x
dpeending on case and cpu etc. - this is ONLY for downscaling of an
image buffer betweeb 50% width and/or height up to 100% of width and
height. it's a special case optimization that cuts down the complexity
of the full super sampling filter to just do a bilinear interpolation
which is actually strictly correct for this size range and shouldn't
drop quality. it uses fixed point (16.16) to do the sup pixel sampling.
no mmx/asse or neon, but we could actually easily use it as we do use
mmx/ee and neon in the bilinear upscaler to do interpolation so this
would work here too. it just requires time and effort to make yet 2x
more special cases and use the ASM to do the hard slog here.
@optimize
we have some duplication of errors between Eina_Error and errno.h,
however we should use Eina_Error to extend the traditional errno.h
system.
then change eina_error_msg_register() and
eina_error_msg_static_register() to return a magic bit to state the
number was registered, and on other functions test this bit in order
to operate on registered values, otherwise fallback to errno.h, such
as strerror().
It also deprecates 2 clear duplicated errors:
- EINA_ERROR_OUT_OF_MEMORY -> ENOMEM
- EINA_ERROR_TIMEOUT -> ETIMEDOUT
There are two details when using strerror():
- old behavior did not return strings for non-error, such as
"Success" or "Unknown error ${N}"
- thread-safety issues: since we must be thread safe, then use
strerror_r() and eina_stringshare_add() that value, keeping a hash
of cached values
Summary:
Visual position of ellipsis item should be set according to
its bidi direction. But, by setting visual position in same way
as logical position, the end ellipsis could be put opposite side.
Also, start ellipsis must placed on left side of RTL text.
@fix T3187
Test Plan: Test an sample on T3187
Reviewers: tasn, woohyun, herdsman
Subscribers: raster, Blackmole, z-wony, cedric, jpeg, minudf
Maniphest Tasks: T3187
Differential Revision: https://phab.enlightenment.org/D3769
Summary:
Image files can have orientation information. Elm Entry have to
call evas_object_image_load_orientation_set() for showing image
with proper orientation.
Test Plan: N/A
Reviewers: herdsman, raster, jpeg
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D4244
Summary:
We can't assume the given font is same with current fdesc by comparing string.
Since Evas starts to supporting "auto" for language,
the given font string should be parsed once before comparing it.
@fix
Test Plan: N/A
Reviewers: tasn, raster, cedric, herdsman
Subscribers: jpeg, minudf, z-wony, Blackmole, woohyun
Differential Revision: https://phab.enlightenment.org/D4227
this fixes a seemingly small typo that would only turn up with fixed
point, but not floats as we have by default whwere we had
x = 999;
instead of
x = FROM_INT(999);
shouldn't be visibule unless you disable float support and use fixed
point in edje.
it was silently ignoring the parent, which led me to hours trying to
investigate why my code wasn't working just to realize my parent
wasn't being assigned and refcounts were screwed.
Summary:
Regardless of the password mode of Entry,
_edje_entry_imf_retrieve_surrounding_cb alwalys passes plain_text
to any callers who register that callback.
This commit replace plain text with '*'
because current behavior could be a security hole in some case.
Reviewers: woohyun, id213sin, jihoon
Reviewed By: jihoon
Subscribers: cedric, jsuya, z-wony, jpeg
Differential Revision: https://phab.enlightenment.org/D4238
Only perform the single value/return type substitution on properties if the
void return type is implicit (i.e. NULL return from function_return_type_get),
following the eolian-C implementation as we use the generated headers.
Also update example after Eo-Efl changes.
Summary:
If native surface is multiple buffer pixmap such as named pixmap, Evas should recreate eglImage everyframe.
Because DDK get the buffer ID once at eglImageCreate time.
So if internal buffer ID is changed, should recreate eglImage.
Test Plan: Wearable Tizen.
Reviewers: wonsik, raster, cedric, jpeg
Reviewed By: jpeg
Subscribers: spacegrapher, dkdk, cedric
Differential Revision: https://phab.enlightenment.org/D4211
The previous commit exposed an issue with the region test
does not take into account the scale down factor.
Not a @fix in itself, as it depends on the previous patch.
Summary:
If both region and scale_down has set, ERR would be returned by loader of jpeg.
@fix
Test Plan: sample code
Reviewers: raster, jypark, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4200
I have no idea how photocam even worked before. @jiin.moon
exposed a major issue in photocam by fixing how the jpg loader
scale down + region load combination works.
@fix
Since the new event type is the same for multi and single
touch (finger id is the only difference), multi touch listeners
would get triggered even for single mouse events.
Fixes https://phab.enlightenment.org/T4345
Fixes https://phab.enlightenment.org/T4378
This raises a question though: How do we want to differentiate
between single & multi touch events in the eo interfaces?
Rename the type to something more sensible and change it to remove the
last remanent of Eo1. This fixes a fixme that has been there for a
while.
The type doesn't really matter, it just looks nicer with the va_list.
This commit implements a sort of CoW for the vtables. The vtables are
usually just linked to and refcounted. When we need to change them we
allocate new ones and copy them over so we can write to them.
I wrote some code to measure the effectiveness of this change. When
running elementary_test (and immediately exiting) I saw that out of the
total number of vtable chains (561) that were needed by the classes in
the EFL, 79 (14.08%) were reused. Considering that I had to add
refcounting (unsigned short, but let's consider it's the size of a word
because of alignment), I would calculate the saving as such (in bytes):
Number of items in a chain (refcounted block): 32
32 bit:
sizeof(chain_node) = 8
Mem wasted on refcounting: 561 * 4 = 2244
Mem saved because of sharing: 79 * (32 * 8) = 20224
Total save: 17980 bytes
64 bit:
sizeof(chain_node) = 16
Mem wasted on refcounting: 561 * 8 = 4488
Mem saved because of sharing: 79 * (32 * 16) = 40448
Total save: 35960 bytes
Wow, we use a lot of memory in Eo classes, I'm sure we can
save even more if we put our hearts into it (change the shareable units
to be smaller to increase the chance of sharing).
This is internal and doesn't affect API/ABI so we can change this even
further with time.
This also improves efl_object_override(). This should now be quite
memory efficient (don't abuse, but it's not a big hogg as it was), so
feel free to abuse that one and rely on it in API.
@feature
The previous patch 1185c40e50 shows how having
two event types for the same thing (key or mouse input events)
was a bad idea. The only guaranteed order of callbacks is the
priority but even that was not enforced (since legacy or eo
style event had to come first).
Fixes T4310
Previously, it was limited to the part block. It was a mistake
and should have been put in the description instead, as this
allows it to change during state transition (like visible, or
proxy_src_visible).
@feature
This macro is behind a BETA ifdef. Shouldn't be used here,
unless we want to remove the beta flag and stabilize
efl_build_version_set() now.
Thanks Stefan for the report.
@fix
This places the following behind beta:
- efl_quicklaunch_fallback
- efl_build_version_set
I don't think EFL_MAIN would have worked without BETA API support,
so no need to expose those for now.
@fix
Summary:
When a theme is loaded, Elm.Layout updates strings with new theme.
This adjustment will be performed by Edje. (See D4219)
Reviewers: cedric, jpeg
Subscribers: taxi2se
Differential Revision: https://phab.enlightenment.org/D4220
So it may be used outside EO (eina error is what I have in mind).
I believe it doesn't need to be redefined in all EFL libs, especially
since it's not used on Windows yet.
Summary:
There are 2 ways to toogle the panel, by using elm_panel_toggle API and by interaction such as flick or click on panel.
This toggled event can help us to know the time when the panel toggled by all case.
Reviewers: woohyun, kimcinoo, Hermet, jypark
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4180
Summary:
Using edc script, there is no way to know evas object is set on swallow part or not.
So, to know that, some elm widget send signal and save some information to edc variables when object is set on swallow part.
I think it is helpful to make better edc script.
@feature
Reviewers: herb, cedric, raster, id213sin, jpeg
Reviewed By: jpeg
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D4176
- convert methods to property setter/getter
- remove "values" block when getter returns read_only value
- fit the type of params of eo funcs to those of legacy APIs
The new event info being an eo object, it needs to be properly
cleaned. I believe this memleak would not show up much in massif
or other checking tools, as it's a stringshare leak. Also, most
of those event objects will in fact be reused and never deleted.
@fix (backported as c58e293d34)
Summary:
Main trouble in handle click and double click event.
If user hold Esc and do double click on genlist item,
it still highlighted but not selected, because between
click and double click genlist process a key down event.
This patch add check the item state on mouse up, check
a item state - if item highlighted but not selected,
unhighlight it.
@fix
Fixes T3986
Test Plan:
1. run elementary_test Genlist
2. hold ESC
3. double-click on few genlist items
Reviewers: raster, cedric, NikaWhite
Subscribers: seoz, jpeg
Maniphest Tasks: T3986
Differential Revision: https://phab.enlightenment.org/D4222
handle smart member add if the object is invalid withotu crashing.
also don't abort and crash with other invalid states like of smart
object or child is to be deleted, has no layer etc. - continue on
safely without a crash.
@fix
thsi fixes invalid memory access to already freed memory in a parent
call where a child freed it by reference counting the struct to
ensuire it stays alive in the parent func using it.
@fix
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
fill spread and angle have never been used - resevred intitially for
gradient objects, but never put into use, so remove to get rid of
junk/memory footprint etc.
@optimize
edje was keeping every edje object created in an eina list so it could
access them later. not really great when every list node contains at
least 4 pointers (data, next, prev and accounting, possibly magic
too). also ever time an edje object is deleted it has to remove from
this list which means... walking the list to where the obj is... not
great. replace with an inlist which is just 3 ptrs, no extra pressure
on list pool and removal os O(1) too.
@optimize
Clang raised the warning:
redefinition of typedef 'Efl_Object' is a
C11 feature [-Wtypedef-redefinition]
for every compiling unit including Eo.h, which
caused a huge console pollution during compilation.
EAPI elm_win_type_set
EAPI elm_win_name_set
Those two APIs should never have been part of the legacy API,
but they have been generated since at least 1.16. The commits
1aceb3bc19
and
41aa19447c
removed the legacy symbols generation. It seemed like a good
idea since the APIs shouldn't exist, but in fact this broke
ABI. I hate this. So sorry about it.
I'm adding them back in with no documentation and as
EINA_DEPRECATED.
This is an emergency commit before the 1.18.x release
announcement.
Fixes T4344
@fix
if clipped color changed and not object color on image objects, then
the updates could be missed. this also fixes some corner cases where
opaqueness should not count for update deletion too. this bug seems to
have been here a while unnoticed.
fixes T4246
@fix
This reverts commit d19cd4e63c.
This causes a SIGBUS error on OpenBSD when closing any application. As this
was safety patch only I will revert it for the 1.18 release and we can work
out what breaks OpenBSD here for 1.19
Fixes T4332
alpha
If we have a borderless alpha window, then make call to set opaque
region accordingly. This still avoids calling opaque_region_set twice,
yet maintains opaque region for the above mentioned case.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
combobox: add Multiple selection feature
This exposes problems with name clashes and behaviour change for the legacy API.
Revert for now and we can revisit for 1.19.
Git revert got to confused as this was committed before we merged elm in. I did
the revert manually.
When trying to create a window, the WL EGL engine creates
an ecore_evas and connects to the wayland display. But if
EGL is not supported (in weston with nvidia for instance),
the egl initialization fails and the window must be detroyed
in order to fallback to wayland_shm.
This led to a double disconnect from the wayland display
as both the ecore_evas del and the error handling code
were trying to disconnect.
Also, use ref == 0 in two places rather than ref <= 0, as
it can prevent double frees in bad situations (ref < 0).
And reset a global variable to NULL on shutdown.
This affects pointer & key events.
Here's the scenario: in Enventor, an elm_button object
grabs the up & down keys for the autocompletion feature.
But the entry will also listen to up & down to move the
cursor. Moreover, the scroller listens to up & down arrow
keys to scroll around.
So, when calling the new eo event first, the scroller
or the entry would catch it and do some action. In
Enventor, this triggered a hide on the magic anchor
object. Bad, bad.
I fear something will be wrong with this order as well.
Both events (legacy & eo) should be called simultaneously,
following the order & priorities that were set.
This being said, in real life most of EFL internals will
move to the new eo events (right now elm relies on the
new events, but edje doesn't, yet), so any app grabbing a
key probably wants to handle the event before elm. Same
for an app using focus manually.
Fixes T4310
Due to the way the clip geometry calculation changed
(see 25d77bc1d2) to be
based on the bounding box rather than the raw geometry
of smart objects, the internal function
evas_object_is_in_output_rect() now returns true if the
rectangle intersects with that bounding box, even if
it's outside the raw geometry.
This breaks the drop area in E's pager, as it relies on
evas_objects_at_xy_get() to find which objects are at
this point. What I saw on my desktop was that only the
lower 10 or 20 pixels were droppable in the pager,
as maximized windows would have shadows covering the
upper 20 pixels or so.
Arguably objects_at_xy_get could also return all objects
at (x,y) including smart objects that extend beyong their
geometry. This can be added as a flag in the EO API, but
not in the legacy API.
while removing children from a parent smart object, the destruction of
the smart obj encountered childrne with NULL parents. they were in the
child list but had no parent. This was totally odd and unexpected,
thus caused an infinite loop trying to dlete a child that won't be
removed from the list because parent is NULL thus it cna't find the
parent list to remove it from. This works around that and then
complains with an error. The workaround also seems to have encountered
what might be a compiler bug so I prepended to the layer object list
rather than appended. this at leats stops the hang.
@fix
We use disabled across our APIs so better fix this new one to unify it before
we release. This was introduced in commit 0116e07288
Many thanks to DaveMDS for reporting this!
After my input event changes, the propagation path has been altered
from:
ecore_x -> evas_event
to:
ecore_x -> ecore_input_evas -> evas_event
But ecore_input_evas was ignoring cancel events by default. There
should be no good reason to disable cancel events anymore,
according to @jypark. Also, this fixes an actual bug :)
Fixes T4301
Without this I get errors from the clang compiler used on Travis for some OSX
builds:
../src/lib/eina/eina_inline_lock_posix.x:845:27: error: use of undeclared identifier 'getpid'
The specific handling for (0.0 <= ellip < 1) doesn't support multi-line cases.
One of the main reasons is that we haven't had the chance to define the wanted
behavior for multi-line.
This is a temporary hack to fix a segfault. The behavior is still undefined,
though.
Fixes T3885.
this addresses a path that could be too small (unlikely due to max
path generally being 4k), for a stack buffer. this should make
coverity happy. fix CID 1356634
we don't give any debug or info if a module in a list fails to load.
we should at least offer debug info. do that. this fixes coverity
issue CID 1039687
as accoding to coverity CID 1353593 chekc returns of the above
function and return/continue as apporpriate if these fail (which
generallly they shouldnt unless some end of the dbus client is broken).
so vpath has a fallback if all things bad go wrong to mkdir a homedir
in /tmp and if that fails use /tmp or use / - if $HOME isnt set ... but
$HOME alwasy should be, so this is only for "a broken system".
this should fix CID 1354286
fix CID 1360466
we don't actually chekc if efreet inits right with elm_need_efreet()
as we just assume it inits (And there pretty much is no case where it
will not), but make coverity happy and check.
@fix
Summary:
When map source is set by elm_map_source_set(),
it foreaches enlisted sources, and should return
when the source cannot be found.
@fix
Test Plan:
Call elm_map_source_set(map, [any type], [Wrong source name]);
Observe ERR("source name (%s) is not found", source_name) message.
Reviewers: Hermet, cedric, jpeg
Reviewed By: jpeg
Subscribers: conr2d
Differential Revision: https://phab.enlightenment.org/D4212
if the text didnt actually change we generat a lot of noise in changed
signals for no change at all in text. shortcut this and check the new
and old text and compare ptrs, 0 length and strings etc.
this fixes T4045
@fix
fixes CID 1361219
if you had an env var a few gb in size or we had stuff in the last bit
of memory address space this might be an issue, but that won't happen.
@fix
fixes CID 1361220
in theory yes end minus start could be insanely huge or end be very
high in memory thus causing an overflow. this would have to be in the
last few bytes of memory space, so it never going to happen. and the
input from the env var has to be sane anyway as its user controlled.
@fix
This reverts commit 8a988717e1.
It seems we can't actually inherit from class more than once and neither eo
nor eolian will complain about it. You will just get random weird behavior.
This patch should come back once we have made an interface of edje.
so ecore-con already cleared the slaves list on shutdown but an
ecore_thread may stillbe active and thus removing from a NULL slaves
list wont work, so skip if slaves is NULL.
@fix
the way eina does sempahores, they can NEVER be sensibly shared
cross-process portably. so enabling sharing is a pointless idea. in
fact some os's like openbsd check if the sempahore addres is in a
sharable mem region and deny init if it is not. on osx you dont use
shared memory but a sempahore name you share instead... and this is
not exposed thus it can't be shared either. if we did process sharable
semaphores we'd make shm segments and/or name them in a sharable way
were you can share the idenitifer of the shm segment and/or the offset
address or name from osx. but we don't, so making them
process-private is the right thing. sharable sempahores will need a
whole new api.
this also fixes osx naming to make the name pretty unguessable/private
and opened exclusive (or it fails) by using pid, sem counter, and 4
random numbers. it's not a security mechanism as the create will fail
if there is a clash. chances are low. we unlink before anyway. good
enough for osx for now.
@fix
There is a line to check select_mode in _item_select function.
The item selected(clicked) callback is called when the select_mode is ELM_OBJECT_SELECT_MODE_NONE.
The checking select_mode should be handled at one place.
this should make ecore_imf testable with empty env vars also meaning
no env var and the make check test will now ensuree this is set to
exactly test this.
This is to avoid ERR messages such as "object is not an image!"
that can happen when a legacy image function is called on an
eo image (proxy, 3d, ...) object.
Note: This does not fix any known issue, but it's probably a safer
solution.
This removes some useless code in various places, where the
switch from eo_do() to standard function call was not properly
refactored.
This changes:
type ret = 0;
ret = my_eo_function();
return ret;
To:
return my_eo_function();
In elm_image, efl_file_get() was called on a legacy Evas Image
object, which generated an ERR message. This was way too much
noise for something that actually works.
It's best not to call efl_file_get/set API on a legacy
object, but it can be convenient for smooth code transitions
from legacy to pure eo.
Also, add safety to those APIs.
Test case: elm_test genlist dnd, start dragging anything.
this fixes errors like:
ERR<4864>:eet lib/eet/eet_lib.c:645 eet_shutdown() File '/home/osauser/.cache/efreet/icons___efreet_fallback_localhost.localdomain.eet' is still open 1 times !
@fix
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.
Because of the translation between legacy and eo key
events, the on_hold flag state could sometimes be lost.
Symptom:
elementary_test -to Focus
Click on a button inside the scroller
Up/down would jump to the end of the scroller, rather than
the next button.
Thanks a lot to Woohyeon for the very good bug report.
in 14ec825aa9 cedric added trying
ELM_DATA_DIR env var explicitly for looking for themes. first if this
is NULL then we look in a bogus directory (null). secondly eina_prefix
already takes care of env vars like thsi as part of figuring out
prefix with correct namespace etc. so this just is a bug as opposed to
anything useful and the CORRECt implementation is already done in
eina_prefix.
@fix
This was code for sndfile. sndfile module should have been
used instead of the pulseaudio one.
It led to a build break when having sndfile but not pulseaudio.
Ref: 879d93377b
If the first output we test is disconnected but has the crtc of
the primary display assigned then we'll fail to assign the crtc
to the connector it's really connected to later.
This is a quick hack as trying to sort this out properly may be
too invasive for the upcoming 1.18 and isn't really important until
1.19 supports multihead anyway.
unswallowing set swallowed_object to NULL inside of
_edje_real_part_swallow_clear
so it is impossible to delete object anymore.
And so, changing sources leaves fantoms and undeleted objects.
@fix
This is a corner case bug I spontaneously found.
* Scenario.
A. Disable A widget.
B. Add a child B widget to A.
C. Now B Widget theme will be followed to A that is performed by
elm_widget_theme_apply()
D. This elm_widget_theme_apply() calls elm_widget_disabled_set() (originally.)
E. Now B widget will be logically disabled.
D. Let's enable A widget again.
E. After going through widget disabled sequence, elm_widget_disabled_eval()
will be called in the last
F. In this function, A widget tries to enable its children. But B widget won't
be enabled because its logically disabled!
Acutally, nowhere widget change children's disabled states logically,
but it propagates its state to children within volatile way so that
A widget perfectly keeps the disabled/enabled state with its children and
recover the children's enable/disable state once their relationship is cut off.
@fix
now set default wheel scroll accel to 7 as epr the confgi upgrade code
and also improve the acceleration to "rely on itself" and feedback
loop. this is so much better now i can scroll through things in a
whisk with the wheel.
fixes T4263
@fix
so drop trying to appease the openbsd packages and stick to "upstream
so major versions" and let users fix their systems with symlinks. also
report what we are looking for so they have a chance to symlink to
make efl happy.
at some point we should make a single simple runtime lib linker
subsystem in efl so all these errors are reported in the same way,
input libray names are listed in a simple consistent way etc. etc.
for now we have 3 locations in efl that do this and they are roughly
similar. we can unify it later.
Otherwise *.eo.h files will be installed twice, so it can break a
parallel install on fast machines, whereby two make jobs may run
concurently, trying to install the same files.
With elementary 1.17.1:
/usr/bin/install -c -m 644 elm_sys_notify.eo.h
/usr/bin/install -c -m 644 elm_sys_notify.eo.h
usr/bin/install: cannot create regular file 'elm_sys_notify.eo.h': File exists
Fixed in Buildroot:
https://git.busybox.net/buildroot/commit/?id=bba82cbfe4caaa01366fc74fb3b6f4c41587a28b
While applying the fix on latest master (1.18.0-beta1), the only
remaining *.eo.h file is efl_ui_text.eo.h.
Fixes:
http://autobuild.buildroot.net/results/7e9/7e9caffe442dcf40ee41da56fd3f55ffedf3a291
Signed-off-by: Romain Naour <romain.naour@gmail.com>
eocre-imf loaded all input modules even though an env var was set to
tell it to use a specific one. this just wastes memory and slows down
startup time. this uses the env var to choose ro auto-guesses wayland
or xim input if the right env vars are set, and if it doesnt know
falls back to "load all" as before. this saves 28kb of dirty pages.
this helps address T4227
@fix
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
so libpuls and libsndfile suck in dependencies. they suck in so much
that by the time linking is done we've written to about 230kb of
PRIVATE MEMORY as dirty pages in symbol tablesm global veriables etc.
etc. - this is just horrible. especially if an app never makes any
sound... it's just wasted memory. this stuff is invisible to normal
memory debug tools. so this begins to address things. please see
T4227. my numbers now put me at:
1780Kb total dirty writable mapped from library file pages. down from
2012Kb.
This fixes some memory bloat reported in the above ticket, but there
is more to fix for sure.
@fix
lib/ecore_x/xcb/ecore_xcb_icccm.c: In function ‘ecore_x_icccm_name_class_set’:
lib/ecore_x/xcb/ecore_xcb_icccm.c:320:11: warning: ‘length_name’ may be used uninitialized in this
function [-Wmaybe-uninitialized]
s += length_name + 1;
Looking at the code this is indeed possible so better play safe here.
Summary:
There is no "ELM_MULTIBUTTONENTRY_EVENT_CHANGED"
Because of this, MBE's Internal entry did not get a changed callback.
And MBE does not have changed smart callback.
Those things made by someone during EO interface working i think.
Test Plan: elementary_test
Reviewers: Hermet, cedric, jpeg
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4195
Summary: it doesn't need to call elm_rescale() in _config_flush_get() when the scale value is not changed.
Reviewers: jaehwan, cedric, raster
Subscribers: minkyu
Differential Revision: https://phab.enlightenment.org/D4186
Summary:
There are critical memory-leak in filter queue process.
It looks typo so I fix it and refactoring queue iteration logic to spend
less time in for and while loop.
Test Plan: Tested by filter elementary test
Reviewers: shashank0990
Reviewed By: shashank0990
Subscribers: raster, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4170
ok. so here's the issue at least now. we have eo objects in the canvas
and they have a refcount of 2 user_refcount is 0. the calls stack does
NOT show we are calling callbacks at that time on these objects. they
are not in the backtrace (the canvas is, the objects themselves are
not).
SOMETHING is keeping 2 eo "internal" refs on these objects and i have
no idea what/how/who. it's a royal pain in the butt to find out as the
only way is lots and lots of logging and you get drowned in the
logging...
so what I have now done is a super ugly workaround that detects these
zombie objects that refuse to die and just FORCES them to die when the
evas canvas frees and clears out layers.
ac10a00acc doesn't really cause the
issue, it just brings it out in the open for all to see far more
easily. but something is deeply wrong SOMEWHERE with SOME objects and
our refcounts.
this fixes T4187
Summary:
In some use cases, font (style) of elm.text in elm entry is changed
on focus and unfocus. user must send "size,eval",elm signal to entry to
force recalc, but this was not happening. This patch fixes that issue.
fixes T4233
Reviewers: tasn, herdsman, raster
Subscribers: cedric, seoz, jpeg
Maniphest Tasks: T4233
Differential Revision: https://phab.enlightenment.org/D4196
lots of unchecked malloc/calloc returns, so check these, unwind safely
and complain (highly unlikely to happen unless the system is about to
fall over). this should fix T4230
Fix legacy functions elm_fileselector_path_get and
elm_fileselector_path_set.
Make these functions call the correct function when called with a
specialization of Elm.Fileselector (instead of the base class function).
Create internal versions of path set/get functions to avoid warnings
about deprecated function calls.
Fix T4198
@fix
As it turns out, when we vt-switch Back to a running Enlightenment,
it ends up processing all outputs and causes Evas to create a giant
canvas to cover all outputs. This new canvas causes a blank screen as
the pageflipping fails (due to fb being too large). As we do not yet
support output on more an one screen anyway, temporarily disable any
output which is not marked as primary.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This commit adds a function we can call from inside Enlightenment in
order to fix an issue with VT-Switching. The problem was when a client
application is open, the compositor does not pass along the key-combo
for vt-switch but rather sends it to an application. This patch (along
with upcoming Enlightenment one) fixes the issue.
NB: This late API add approved by Stefan ;)
@feature
Signed-off-by: Chris Michael <cp.michael@samsung.com>
When we are calculating minimum size hints for the window object, we
should be adding framespace to maximum size, not subtracting it...
Honestly this just looks like a brain-dead typo.
Fixes T44225
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Previously, if the window had no frame object, then this could would
have been setting opaque region twice (for the alpha case). We can
avoid this if we move the alpha check to below the frame check.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Problem: crash in assert() in terminology.
Scenario:
Open Terminology,
Split V by keyboard,
Move mouse to split 2,
Create tab by keyboard
--> abort() in terminology
Cause:
An extra mouse,in event happens during edje_object_unswallow
inside elm_layout_content_unset.
Root cause:
efl_part() in elm_layout had a side effect: edje_recalc on the
edje object. Causing its geometry to be "properly" defined and
the mouse event to trigger callbacks.
Solution:
Avoid calling recalc... somehow.
Conclusion:
Without adding any new API, edje edit provides internally the
information that we want: type of an edje part (for box & table).
Fixes T4221
See T4028
See T3509
so e is being stupid and creating an ecore-x image forevery single
window if in x11 mode if it needs it or not. this results in having ti
allocate an actual x image and shm segments. work around this and get
bit order from somewhere else than the x image itself thus avoiding
the allocation until a real get or put is done.
@fix
Since session recovery can cause apps to constantly try and reconnect
to the display server, we probably should not be filling logs here. On
the other hand, any failures on first connect are now not going to
display any error messages.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
What a mess... Assuming efl_part() had no side effect on the
object it took me hours to figure out that there was a wrong
call to _edje_recalc_do in the efl_part() function itself.
That was bad, and existed because efl_part() used to be
efl_content_get().
efl_part() should not have any side effect.
Also, fix a return value in content_remove.
Fixes T4214 (invalid redraw and crash in terminology).
this PARTLY addresses T3556 ... this handles our own conversion of
makrup to string then to a windows newlined string. this half. i wrote
the conversion code for the other way but it's unused as i am unsure
where exactly to plug in it. following the cnp code makes me not sure
where it goes so for now - not there, but ready to go.
to be clear. out API is unix text. utf8 strings at our api with UNIX
newlines. that is our api. that is the text we accept and produce. if
you deal with another file or interface that does not provide this
then the job of conversion is AT THAT POINT. eg elm_cnp.c has to do
this. as would file loads of text files (and saves) etc. - anything
else like forgivingly handling anything at the api level makes it
totally unclear what our api is and what should go in and come out.
to be portable we have to define what it is and the most portable
thing to do is at the api level within a process we define one and
only one format. UNIX \n format.
@fix - partial
This fixes the following error message:
Object '0x40000002f0000030' still has a parent at the time of destruction.
Though I actually wonder why ref == 0 and there was a parent.
Something is still a bit fishy here.
so our sysv shm segments were both over-permissive (nothing bad
really, just other users could read and write to/from our pixel data
destined for the screen... they could do this to x11 directly anyway
so no real issue), but be more restrictive and use 0600 as xserver
runs as root so can read/write anyway and we only want our own uid
access. but even more - fix our shm segment flushing to not keep lots
of segments floating about like a bad smell when we don't need them.
we had a cache but it wasnt flushed when it should be since async
rendering turned up. this fixes that and we're back to agressively
flushing them out when idle.
@fix
Summary:
- In the callback for ELM_PLUG_EVENT_IMAGE_DELETED,
it tries to get elm_conform's data from NULL.
@fix
Reviewers: Hermet, woohyun, cedric, jpeg
Reviewed By: jpeg
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4177
is_visible returns 0 if no_render is true, so should
was_visible as well. Yeah, there will be problems if
no_render changes on the fly. Don't do that.
Fixes T4193
This is annoying to see key events flooding a terminal ... so comment
out this debug statement. If a developer really needs this, then
enable it locally.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This reverts commit 57d77c9041.
We are in freeze right now and this is a new feature. Considering the access
API is still in beta I might have let this slip by but its does not even build
../../../src/tests/elementary/elm_test_popup.c:27:1: error: conflicting types for ‘elm_object_part_access_object_get’
START_TEST (elm_object_part_access_object_get)
^
In file included from ../../../src/lib/elementary/Elementary.h:240:0,
from ../../../src/tests/elementary/elm_test_popup.c:6:
../../../src/lib/elementary/elm_object.h:521:54: note: previous declaration of ‘elm_object_part_access_object_get’ was here
EAPI Evas_Object *elm_object_part_access_object_get(const Evas_Object *obj, const char *part);
^
../../../src/tests/elementary/elm_test_popup.c: In function ‘elm_object_part_access_object_get’:
../../../src/tests/elementary/elm_test_popup.c:39:47: warning: passing argument 1 of ‘elm_object_part_access_object_get’ makes integer from pointer without a cast [-Wint-conversion]
access = elm_object_part_access_object_get(popup, "access.title");
^
../../../src/tests/elementary/elm_test_popup.c:27:1: note: expected ‘int’ but argument is of type ‘Evas_Object * {aka struct _Eo_Opaque *}’
START_TEST (elm_object_part_access_object_get)
^
../../../src/tests/elementary/elm_test_popup.c:39:13: error: too many arguments to function ‘elm_object_part_access_object_get’
access = elm_object_part_access_object_get(popup, "access.title");
^
../../../src/tests/elementary/elm_test_popup.c:27:1: note: declared here
START_TEST (elm_object_part_access_object_get)
^
../../../src/tests/elementary/elm_test_popup.c:39:11: error: void value not ignored as it ought to be
access = elm_object_part_access_object_get(popup, "access.title");
^
Makefile:43996: recipe for target 'tests/elementary/tests_elementary_elm_suite-elm_test_popup.o' failed
make[5]: *** [tests/elementary/tests_elementary_elm_suite-elm_test_popup.o] Error 1
If you still want this in make sure it actually can compile the tests you added.
When we are adding new input devices, if we already have an existing
one (keyboard, pointer, touch, etc) then we should just be
incrementing the device count and returning from the init function.
Previous code here would increment the device count, then continue on
and create a new internal representation of the device. This was
causing issues when a device gets removed.
Fixes T4192
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
so if an img obj had some update regions and was opaque, it didnt del
an update region first before adding the updates, thus possibly
letting update regions pass through the opaque image area
@fix
This commit introduces a new state per input, this state is just for the
keymap and no modifieres, the state is used to get the keysym from a
keycode. This is done to stay compatible with x since keyname in efl is
the characters which are describing the key which is pressed, without
the transformation of shift etc.
This also escapes keycodes from beeing a utf8 character, and passes out
a keyname instead of the utf8 symbol.
fix T3976
fix T4184
In all code paths leading to this goto statement being called,
'iter_array' variable will already be NULL, so checking it's validity
is just logically dead code.
Fixes Coverity CID1357392
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that eina_safepointer_get returns a NULL promise here
(checked 20 out of 21 times). As eina_safepointer_get can return NULL,
we should check the validity of 'promise' here before trying to
derefernce it later.
Fixes Coverity CID1356625
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that this is logically dead code. As we check the
validity of 'model' above this function, then testing 'model' for
validity again in this expression is just logically dead.
Fixes Coverity CID1356617
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that this is logically dead code. As eo_item has been
dereferenced on all paths leading here, and it is also set above, then
it cannot be NULL at this point, thus this check is logically dead.
Fixes Covierty CID1355585
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that we do not check the return value of
_eldbus_model_proxy_load here. We likely should be checking that in
case the model proxy fails to load so we can return proper values here.
Fixes Coverity CID1355233
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that we call _circular_dependency_find here without
checking the return value
Fixes Coverity CID1349867
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Coverity reports that we are missing breaks in the switch blocks here,
so add missing breaks.
Fixes Coverity CID1347413 and CID1347414
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
this fixes over-rendering (where we render the whole image) if it has
a border set. do proper region updates if the image has a border but
is not scaled and the fill is at 0 0 and sams size as the object etc.
etc. so this is simple and doable.
this fixes T4123
this just makes it more obvious where its handling the opaque middle
fill case. see T4123 - i was just reding the code and this seemingly
does the right thing.
This reverts commit fb24955782.
While this patch did fix the issue of toggling window Borderless state
in EFl-Wl client applications, it is "Enlightenment's" stance that it
will not change to match this fix, so this has to be reverted else
internal E dialogs show up with no border (due to them using SSD).
This is going to re-open T4092 sadly :( and now leave that bug with no
possible fix.
ref T4092
Fixes T4126
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This is a (minor) API & ABI break in Eo.h!
I say minor as eo_override shouldn't be used yet (EO is unstable
and this patch includes all the use cases in EFL).
I'm not very happy about the new form of the macro, but it avoids
two things:
- passing in a struct (valid in C, but never used in EFL)
- using a GCC construct to create structs on the fly
It was inspired by the event array define, but I don't think
we need the runtime memcpy here.
See also:
https://gcc.gnu.org/onlinedocs/gcc/Compound-Literals.html
This fixes the following ERR message:
ERR<30400>:eo /home/jpeg/e/core/efl/src/lib/eo/eo.c:462 _eo_call_resolve()
in ../src/lib/evas/canvas/efl_event_input.eo.c:3: func 'efl_event_flags_get'
(1483) could not be resolved for class 'Efl_Event_Hold'.
This is an EO-API only fix.
This fixes the following ERR:
ERR<23969>:eo /home/jpeg/e/core/efl/src/lib/eo/eo.c:462
_eo_call_resolve() in ../src/lib/efl/interfaces/efl_event.eo.c:5:
func 'efl_event_instance_get' (1479) could not be resolved for class 'Efl_Event_Hold'.
i don't know for sure if this fixes T4103 but in theory i think it
might given a reading of the backtrace and a guess at what might
happen, so try this fix. it doesn't hurt and can only help.
@fix
there was a leak of elements on the itrator stack when the stack was
flushed still having some items on it, thus losing their pointers and
never freeing them.
@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>
eo_override would leak the vtable if called multiple times, this
fixes that. Also, it is now possible to revert back to the original
class' vtable by passing in { NULL, 0 }
I believe it is thus possible to incrementally override more
functions on an object. Absolutely not recommended, but should work.
But it is not possible to selectively revert back to the original
class implementation on a single method. Use eo_super for that,
or revert back the entire object overrides.
PS: Is it normal that we pass in a struct? We never do that in EFL...
This removes the need for khronos_[u]int64_t as well as the special
typdef EvasGL[u]int64.
Hopefully this should work on all platforms (note: [u]int64_t is
used in Eina APIs, so it is already required for EFL apps).
Fixes T3200
This should fix T3309
Snapshot objects are image objects, so the function is_opaque
exists. No need to bypass it. Also, alpha rectangles are not
opaque. Assume that anything with a filter is not opaque.
All of this fixes T3309 but the main point was on snapshot
objects (probably because the only point of a snapshot is
to apply a filter on it).
This commit partially fixes a phab ticket where toggling window border
state (elm_test -to "Window States") would cause the frame to be
destroyed thus making the window contents appear to jump around
ref T4092
NB: I call this a 'partial' fix because while it does fix the issue of
efl client apps, there is still some breakage in Enlightenment that
needs investigating.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Previously a drm event could create a new output or disconnect an output,
but could not re-enable an existing output.
This results in failure to properly wake from screen blanker.
Summary:
When using EDC written by another person, developer usually
does not need to know all specified parts in EDC but editable
like swallow, text or textblock parts.
The required flag tells whether the part is intended to be used outside EDC.
Reviewers: jpeg, cedric, raster, Hermet
Reviewed By: jpeg, raster
Subscribers: raster, NikaWhite, kimcinoo, jpeg
Differential Revision: https://phab.enlightenment.org/D3579
We're currently doing screenshots in E under wayland by copying data
out of the framebuffer, mmaping gbm buffers makes screenshots work
again when rendering with GL.
This is essentially a cleaner redo of ef817f15f0.
Logic should be exactly the same as there, the different is that this
one shares the code between OT and non OT.
Please refer to that commit for more information.
This was not done correctly. This split the code, which is essentially
the same for both OT and non OT. It's the same logic with some minor
additions for OT, so most of the code should be together.
This reverts commit ef817f15f0.
Fixes T4068.
Simply querying the last glyph to determine the width of the glyph sequence
won't always work, as OT can have negative offsets (adjusts the placement of a
specific glyph better).
The solution is to calculate the "max width" of some sequence that will
guarantee us proper width results. The worst solution would be to iterate on all
the glyphs and sum up the max width. This is a bit impractical. Instead, we will
inspect just the "cluster" of the last glyph, if one exists.
This should have no performance impact on trivial cases, and very little impact
on the others.
@fix
Summary:
- Previously, mode change, color change updated
sd->selected, but did not send signals to edje of the item.
- Also, mode change in colorselector set sd->selected to
the first item even when palette is visible in previous mode.
- Now, when mode is changed, sd->selected is set NULL and send
unselected signal if palette is invisible, and keep sd->selected
if palette is visible.
- sd->selected is set NULL when color is changed in picker and
color bars because previous selected item is not current color of
colorselector
Test Plan:
1.
1) launch elementary_test colorselector.
2)Select any item and check the item is selected when palette is visible.
3) Change mode to Palette, Both, All, and check selected item is not changed.
(Previously, selected item is changed to the first item of the palette, but previous item was seen as selected.)
2.
1) launch elementary_test colorselector.
2) Select any item and check the item is selected when palette is visible.
3) Change mode to Components, Picker.
4) Change mode to Palette, and check none of items is selected.
(Previously, selected item is changed to the first item of the palette, but previous item was seen as selected.)
3.
1) launch elementary_test colorselector.
2) Select any item and check the item is selected when palette is visible.
3) Change mode to All.
4) Click arrows in Picker and color bars.
5) Observe selected item is not unselected when color is not changed, and selected item is unselected when color is changed.
4.
1) launch elementary_test colorselector.
2) Select any item and check the item is selected when palette is visible.
3) Press direction key and check selected item is changed.
(Previously, selected item's edje is not updated.)
Reviewers: woohyun, Hermet, jpeg, cedric
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4156
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
This commit just readds the layer setting code which made the frame
border be above content. This does make sense in some contexts
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
When we make calls to get an Input device's output name, if the device
does not have an output name it's not actually an error so just return
NULL with less noise
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This patch fixes an issue where if you tried to resize a window to
it's minimum size, the contents of the window would draw outside the
window frame. Basically, when we are setting min/max size hints to the
window object we need to account for framespace.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
I guess the overflow was badly handled. Fixing it by using
explicit int intermediate value.
Fixes CID 1356616 and 1356619:
Operands don't affect result
Logically dead code
realloc() can return NULL if size is 0. It's like free().
So, the usage here is correct, and there are probably no
points to interpolate between anyway. I wonder if there
can be commands without points, though.
Fixes CID 1293004
CID 1355234 Logically dead code
The indicated dead code may have performed some action;
that action will never occur.
In eet_data_descriptor_element_add: Code can never be
reached because of a logical contradiction (CWE-561)
Solution: use explicit range within valid values with <= and >=
rather than excluded values with > and <
This patch fixes an issue where wayland windows would not swallow
clients properly. Basically, we should be swallowing the elm window
layout directly rather than use a rectangle (as was previous). This
commit also waits until the internal elm theme has been applied to the
sd->edje object before adding the wayland window frame.
This commit has the added benefit of being able to remove an additional
Evas_Object from the internal data structure.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Tom Hacohen
Gustavo Sverzut Barbieri
Daniel Hirt
Marcel Hollerbach
Cedric Bail
Michaël Bouchaud (yoz)
Andy Williams
Jiwon Kim
Minkyu Kang
Jee-Yong Um
Carsten Haitzler
Jean-Philippe Andre
Jean Guyomarc'h
Amitesh Singh
WooHyun Jung
Vitor Sousa
Mykyta Biliavskyi
Jean Guyomarc'h
Subodh Kumar
Mike Blumenkrantz
Derek Foreman
Daniel Kolesa
Jihoon Kim
Prateek Thakur
Artem Popov
jiin.moon
Youngbok Shin
Lauro Moura
Minkyoung Kim
Stefan Schmidt
Jee-Yong Um
JEONGHYUN YUN
Jinyong Park
Thiep Ha
Sungmin Kwak
Vyacheslav Reutskiy
Boris Faure
Felipe Magno de Almeida
Chris Michael
Sungtaek Hong
Vincent Torri
Stephen Houston
Shinwoo Kim
Hermet Park
Subhransu Mohanty
Davide Andreoli
Vitalii Vorobiov
Romain Naour
Daniel Hirt
Woochan Lee
Jonghee Choi
SangHyeon Lee
Al Poole
perepelits.m
Ivan Furs