Can be called by efl_ui_widget_scroll_hold_{pop, push} from another widget (ex., efl_ui_slider) without _efl_ui_textbox_scrollable_set() invocation with undefined sd->scroller here as the result. So, check if sd->scroll is true to avoid this.
The same for *_freeze_set.
When we use efl_ui_textbox and efl_ui_slider simultaneously, the slider invokes the general widget's scrolling hold and freeze behavior, which calls of the textbox's scrolling functions with an obvious error: sd->scroller = NULL. So switch off that in the slider and don't notice any lack of functionality.
Saved sd->freeze as a flag for future use.
Test: elementary_test -to Efl.Ui.Slider (before and after).
check adds a NULL and this messes the format warning checks in the
check macro - so disable this warning where this is done - warning
noise we don't need.
you can compile the them with edje_cc -DBLUR_WINLIST and either
-DBLUR_TITLE or -DBLUR_BEHIND options passed to edje_cc ... if you
compile it by hand. either way it makes it easier to see how to do
this for theme makers.
The POSIX path "/home/mydir/../myfile" can't be tested on Windows.
In MSYS2, /home is replaced with the installation path os MSYS2,
that is something like C:\foo\bar (depending on the value of the
path given with the MSYS2 installer). So this can't be tested with
the value "/home/myfile".
luajit seems to be dwindling. it also has bugs on aarch64 which means
packages for efl like i maintain on arch use lua instead of luajit by
default. reflect this as the default choice now.
like a mask of a mask jut didn't work at all - not to mention
hijacking table clippers.... that's just wrong and so dependent on
implementation internals...
due to slightly different optimizations we need a big more delta inthe
pixel compare. do it properly channel by channel too so we know whcih
channel (a, r, g, b) is wrong
The first branch is the POSIX-compliant one and the one that
should always be taken; __USE_MISC is an unrelated glibc-specific
macro so the code was not being used on non-glibc Linux systems
even though it should be.
All these branches are Linux-specific, so there is no legacy
code to worry about, even if it was actually useful at some point
in history.
found by q66 - strlen happende to work as the optimizer could deduce
this was a const value at compile time, but strictly it wasn't so use
sizeof instead.
@fix
Vincent Torri
Carsten Haitzler
Dmitri Chudinov
Ross Vandegrift
q66