when canceling a all_promise it will cancel all futures. When that
happens, and one future is containing a promise, the value is unrolled,
and delivered as "Operation canceled" if this is happening to the last
future in all or any in race, the promise will then free its base ctx
which is already happening due to canceling.
With this this is not happening anymore.
Summary:
Patch from a series of patches to rename EAPI symbols to specific
library DSOs.
EAPI was designed to be able to pass
`__attribute__ ((visibility ("default")))` for symbols with
GCC, which would mean that even if -fvisibility=hidden was used
when compiling the library, the needed symbols would get exported.
MSVC __almost__ works like GCC (or mingw) in which you can
declare everything as export and it will just work (slower, but
it will work). But there's a caveat: global variables will not
work the same way for MSVC, but works for mingw and GCC.
For global variables (as opposed to functions), MSVC requires
correct DSO visibility for MSVC: instead of declaring a symbol as
export for everything, you need to declare it as import when
importing from another DSO and export when defining it locally.
With current EAPI definitions, we get the following example
working in mingw and MSVC (observe it doesn't define any global
variables as exported symbols).
Example 1:
dll1:
```
EAPI void foo(void);
EAPI void bar()
{
foo();
}
```
dll2:
```
EAPI void foo()
{
printf ("foo\n");
}
```
This works fine with API defined as __declspec(dllexport) in both
cases and for gcc defining as
`__atttribute__((visibility("default")))`
However, the following:
Example 2:
dll1:
```
EAPI extern int foo;
EAPI void foobar(void);
EAPI void bar()
{
foo = 5;
foobar();
}
```
dll2:
```
EAPI int foo = 0;
EAPI void foobar()
{
printf ("foo %d\n", foo);
}
```
This will work on mingw but will not work for MSVC. And that's why
EAPI is the only solution that worked for MSVC.
Co-authored-by: João Paulo Taylor Ienczak Zanette <jpaulotiz@gmail.com>
Co-authored-by: Ricardo Campos <ricardo.campos@expertise.dev>
Co-authored-by: Lucas Cavalcante de Sousa <lucks.sousa@gmail.com>
Reviewers: jptiz, lucas, woohyun, vtorri, raster
Reviewed By: jptiz, lucas, vtorri
Subscribers: ProhtMeyhet, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D12188
Summary:
Also replace all mkstemp(s) and mkdtemp with the eina_file functions in
the source
Test Plan: run eina_file test
Reviewers: raster
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D12170
it wasn't being enabled thus missing an entire optimization limiting
stats to "once per frame" etc. to avoid overly syscall+io happy code
from having as much of an impact
@fix
this won't happen as the setup should not have the # of fd's found in
2st pass and 2nd pass for a dirfd change as nothing should interfere.
comment to that effect and add a paranoid change num_closes to the
number found
Summary: on those terminal, stderr is never flushed, so errors messages are not displayed when, for example, the app crashes
Test Plan:
test program :
```#include <Eina.h>
#define ERR(...) EINA_LOG_DOM_ERR(log_dom, __VA_ARGS__)
int main()
{
int log_dom = -1;
eina_init();
log_dom = eina_log_domain_register("eet", EINA_COLOR_CYAN);
if (log_dom < 0)
{
printf("error register\n");
return 0;
}
ERR("an error.");
while (1)
{
}
return 0;
}
```
without the patch, nothing is displayed, even after a Ctrl-C to finish the program
with the patch, the error message is displayed
Reviewers: raster
Reviewed By: raster
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D12058
Summary: I found wrong API reference group name in mapbuf, Evas, Eet, Eina, Eio and fixed them.
Test Plan: API reference documentation modification only
Reviewers: segfaultxavi, stefan_schmidt
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D12052
Without it an expression like !EINA_LIKELY(a && b) expands
!a && b
Reviewed-by: Cedric BAIL <cedric.bail@free.fr>
Reviewed-by: Vincent Torri <vincent.torri@gmail.com>
Differential Revision: https://phab.enlightenment.org/D12041
`strerror_s` is the windows alternative of `strerror_r` used by EFL.
`strerror_s` never return the error code with the message as
`strerror_r` does, because of that, while comparing the first 14
characters of `Unknown error ` to the message from unknown code 4096
(`Unknown error`) they were accusing being different - in UNIX this
works because the message returned is `Unknown error 4096`.
This error was noticeable at `eina_error_test_failures` test case.
This Diff adds the error code to the message in case of an `Unknown
error`, making the windows implementation compliant with UNIX.
Reviewed-by: Vincent Torri <vincent.torri@gmail.com>
Reviewed-by: Wander Lairson Costa <wander.lairson@gmail.com>
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D12033
This commit does two things:
- Tell the compiler that it is unlikely that we need to grow, and that
it is unlikely that data is NULL. Sometimes the if check for data
would get dropped out by the compiler when it can be ensured that it is
!= NULL. However, if we for example efl_add something and eina_push
the result, the condition would not be removed, as there is no assertion
efl_add would be != NULL.
- Do not hide the array assignment in a branch, but make it the default
branch, this way instruction cache caches the correct instruction, as
branch prediction will now hopefully, due to the hinting, take the
correct branch.
While benchmarking this here (simply in elementary_perf), this reduced
pipeline faults in eina_array_push quite a bit. (Btw. it is hard to track
*which* exact calls to eina_array_push do cause that, as mostly this API
gets inlined, so it was easier optimizing that, instead of the method
arround)
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11997
Summary: when eina_strbuf_replace is used by read_only buffer, this will cause segfault (access invalid memory)
Reviewers: cedric
Reviewed By: cedric
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T8757
Differential Revision: https://phab.enlightenment.org/D11989
Summary:
Found by running terminology's tests with UBSAN:
include/eina-1/eina/eina_inline_unicode.x:
runtime error: implicit conversion from type 'char' of value
-62 (8-bit, signed) to type 'unsigned char' changed the value to 194
(8-bit, unsigned)
Reviewers: #reviewers, vtorri
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11972
libraries are split into deps, external deps, and pub deps.
Evas engines are refactored to use the predefined engine deps.
this is preparation work for efl-one.
Reviewed-by: Stefan Schmidt <stefan@datenfreihafen.org>
Differential Revision: https://phab.enlightenment.org/D11806
Add it to subprojects which are not using it and remove and old
ELEMENTARY_BUILD define we no longer use. This allows us to have a
central place in the main meson.build file to set this variable.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Reviewed-by: Vincent Torri <vincent.torri@gmail.com>
Reviewed-by: João Paulo Taylor Ienczak Zanette <joao.tiz@expertisesolutions.com.br>
Differential Revision: https://phab.enlightenment.org/D11854
if try_no_queue is used - its a thread that clearly wantt to be out of
the queue to run on its own so probably wants to wake up accurately
and thus not be a backgroun rpriority task but and urgent one. ensure
we set up priority accordingly as we didn't before. this should fix
scheduling when under load for vsync
@fix
Summary: EAPI must be defined to dllexport when building DLL, and to dllimport when using these DLL. To achieve this, define EFL_BUILD for each library and module, and set DLL_EXPORT unconditionally. Static library are and will be not supported
Test Plan: compilation
Reviewers: zmike, raster, jptiz
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11834
so i've moved all systemd and elogind support to be runtime only with
dlopen (eina_module) of libsystemd.so.0 (or libelogind.so.0 for elput)
and finding of symbols manually at runtime (if the right code paths or
env vars are set), thus remvoing the need to decide at compile time if
efl needs systemd support or not as it no longer needs systemd
headers/libs at compile time and just at runtime. this simplifies
building a bit and makes efl more adaptive to the final target system
at runtime.
Summary:
void* is an invalid type for calculating pointer offsets, so ensure that
this is always cast to something else (e.g., char*) in the few cases
where it's been misused
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D11816