A long story in a few words: sometimes on macos modules will be compiled
into .dylib, sometimes, into .so suffix. We did not set the suffix
everywhere in our meson build instructions, hence our suffixes have been
differently, which resulted in random load fails on different maschines.
With this commit, we ensure that we suffix all modules the same way.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11650
osx doesn't ship a pkgconfig file for its own zlib, so now we're shipping
a pkgconfig file for their zlib to avoid future issues
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11661
debug name could be called during any time, even before the constructor,
of this super class is hit. So cur can be NULL here.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11659
Summary:
do not take over complete vtables.
This might fix wrong vtable settings when a type is not in the
inheritance of another type, but the function is implemented.
Reviewers: woohyun
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11657
these did not even look at aspect hints when calculating sizing. that
means any attempt to set them would lead to... nothing useful. this
handles horiz/vert/both cases (as best as is possible).
@fix
This reverts previous commit and fixes it in the box layout to respect
aspect in elm boxes. note - this probably needs doing in other
containers too like table...
Revert "elm icon/image efl ui image - respect aspect hints at all if set"
these did not even look at aspect hints when calculating sizing. that
means any attempt to set them would lead to... nothing useful. this
handles horiz/vert/both cases (as best as is possible).
@fix
since paste is now working when called from user source code, this test case can be added
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11625
previously, textpath delays the map calculation to avoid duplicated jobs.
some cases, this job could be delayed to the next frame that occured a
wrong frame result.
This render_pre event gurantees the textpath to update frames exactly.
@fix
'ctx' can be leaked in some case.
this pathc fixes it.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11368
memory is duplicated in eina_content_new.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11641
this shouldn't trigger any canvas changes since nothing is happening
during the freeze
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11631
this can be useful when a test needs to immediately verify whether a render
is occurring
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11630
previously this would always queue a recalc when calling thaw even if the
object hadn't changed
also mimic edje internal behavior with unsetting 'frozen' during force calc
for possible future handling even though it has no effect presently
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11628
we should prefetch the func pointer here.
Later on we are accessing the func pointer in a streak, after that, we
do not use it anymore.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11593
the structure "!strcmp(X, "foo", strlen("foo"))" is equal to
"eina_has_prefix(X, "foo")", and the later is nicer to read, hence this
replaces it.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11620
this is unnecessary:
* for the text object itself, the value is correctly set whenever the
paragraph_direction property is changed
* for inheritance, smart object already implements propagation which triggers
on member add or property change
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11549
this pointer needs to be freed if it's being copied
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11604
Previously we use key strings to detect keyboard shortcuts (for example copy/paste/select_all, .. etc)
Now we will use key code alongside with these key strings (I do not remove the key string, because I am afraid something depends on them because some methods do not send keycode with keys)
This issue can be easily detected when the keyboard layout is not a Latin language, for example Hitting "C" will produce key string depends on related language not letter C, but Keycodes remain the same regardless of keyboard layout.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11606
This removes EINA_UNUSED when obj is actually used
like in a call ed = _edje_fetch(obj);
Please note the macro GET_REAL_PART_ON_FAIL_RETURN
also uses obj.
No functional changes, just cleanup.
Reviewed-by: Christopher Michael <devilhorns@comcast.net>
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11471
This brings into the docs hundreds of methods!
due to the ingroup->defgroup mistake, they were out of any scope
and therefore they were silently ignored by doxygen.
Also, document lots of missing "obj" parameters. Not strictly necessary, but
this further reduces the number of doxygen warnings.
Summary:
The Exactness tool needed usage instructions... and quite some more
fixes. There was copypasta all around.
Depends on D11634
Test Plan: Build and Enjoy
Reviewers: bu5hm4n, stefan_schmidt
Reviewed By: stefan_schmidt
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11637
Since the changes from Marcel to run exactness as LD_PRELOADed library
we can switch back to the normal test.txt file and names.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11636
there came up a issue, where a excatness spawned processes were bringing
up a efreetd instance, when the efreetd instance turned off itself, the
files for exactness were written again, which is wrong. This ensures
that forked instances do not take any actions.
Differential Revision: https://phab.enlightenment.org/D11634
if _src_unit is NULL, the write here would delete the actions, with this
commit we ensure that this is printing an error.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11627
before a few commits, we had the situation that errors were overseen
because the log was simply so big, that errors did not get shown
properly.
With this commit, exactness will simply abort if there is a real issue
in the code.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11624
we should not error when mkdir returns < 0. EEXIST should not result in
the return here.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11618
there is no need to do that, more than that. This is super dangerous,
the display and connection ptr of x are passed from ecore_evas to evas,
if you delete evas before ecore_evas, the later ecore_evas deletion will
destroy the x connection which calls some functions in evas, which is
already freed, which leads to a crash.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11617
this is just a little python script, so you can lunch exactness_play
without the need of handdefining LD_PRELOAD
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11616
this can now be loaded as LD_PRELOAD library
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11615
this refactors everything that is not argc argv related out of the main
method. For later usage
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11614
this seems to be made for compiling binaries before testing. That sounds
like a good idea, however, implementing a full buildtool in exactness is
a bit hard. Hence, using meson for that would be better.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11613
this is just a little python script, so you can lunch exactness_record
without the need of handdefining LD_PRELOAD
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11611
this is now not a binary anymore, that dlopen's a binary, it is now a
library, that can be loaded using LD_PRELOAD. EXACTNESS_DEST is used for
the path of the .exu file. EXACTNESS_FONTS_DIR is used to get the fonts
directory
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11610
all calls taht are not related to env var checking, are moved out of the
main method. That is in preparation for later refactorings.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11623
Coverity reports that 'fd' here is negative, and close() cannot accept
a negative parameter, so add a check to make sure 'fd' is not negative
before passing to close function.
Fixes CID1420318
0 was the wrong seat, and only worked on a small amount of systems
fixes T8639
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11622
_test_name was not used globally, so lets move it to the used scope.
Verbose is not used at all either.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11609
The efl_access_text_attribute_get is resolved by elm_entry.
Please refer to _elm_entry_efl_access_text_attribute_get first.
Uninitialized variables are used for its parameters, and it is able to
return before setting these variables.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D11619
We have them behind an env variable as this is system specific. Enable
them on CI to fully test these paths as well.
While this works out of the box for the native build on Travis it fails
for our docker builds, as the dockerd does not have ipv6 enabled by
default. Only enable for Codecov for now.
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Differential Revision: https://phab.enlightenment.org/D11608