Summary:
after a fork does happen, the new process does not have any self created
threads at all. However, _thread_cb can contain suspend calls of
ecore_thread_main_loop_begin. _ecore_main_call_flush will then wait in
the suspend block for the thread to call ecore_thread_main_loop_end.
However, the thread is dead, the end function will never be called.
Hence we should ensure that we definitly kill every entry in _thread_cb
that has a susped flag on true.
This fixes deadlocks while running the testsuites with
EIO_MONITOR_POLL=1
Depends on D8526
Reviewers: cedric, segfaultxavi, zmike
Reviewed By: zmike
Subscribers: #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8531
Summary:
we need to to this here, otherwise we effectfily start the monitor
later, which means, we can miss a change in the filesystem. However,
this makes things a lot slower.
Reviewers: cedric
Reviewed By: cedric
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8526
Summary:
Little introduction into what eio did before this commit:
Efl.Io.Model creation:
- direct ls of a directory in a thread (A)
- take all the contents of a directory and feed it slowly back into
the mainloop
- when all events have been feeded back to the mainloop and have been
processed: start monitoring (B)
However, any file created between (A) and (B) will not be in the model,
since not the listing nor the monitoring did caputure it. Hence we need
to start monitoring before we actaully start listing. In the callbacks
we then check if we already published something.
ref T7311
Reviewers: zmike, cedric
Reviewed By: cedric
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7311
Differential Revision: https://phab.enlightenment.org/D8525
Summary:
The @ingroup of elm_object_item APIs is Elm_General in elm_object_item.h.
Since legacy prefix of Elm_Widget_Item is elm_object_item,
the prefix of Elm_Widget_Item APIs is elm_object_item.
Consequently, to display the legacy Elm_Widget_Item APIs with
elm_object_item APIs, the @ingroup is changed from Elm_Object_Item_Group
to Elm_General.
Test Plan: make doc and check the documentation
Reviewers: segfaultxavi
Reviewed By: segfaultxavi
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8529
This reverts commit 913a5bff66.
The commit is reverted because the commit implements features during
feature freeze period before release.
913a5bff66 will be restored after release.
This reverts commit 74c48f0296.
The commit is reverted because the commit implements features during
feature freeze period before release.
74c48f0296 will be restored after release.
This reverts commit 30bd541bf2.
The commit is reverted to revert the following commits.
74c48f0296913a5bff66
The above commits will be restored after release and then
30bd541bf2 will be also restored.
This reverts commit d3bb1a7342.
The commit is reverted to revert the following commits.
74c48f0296913a5bff66
The above commits will be restored after release and then
d3bb1a7342 will be also restored.
This reverts commit e3d2a0cf12.
The commit is reverted to revert the following commit.
74c48f0296
The above commit will be restored after release and then
e3d2a0cf12 will be also restored.
There is a wrong case that fails caching ector surfaces.
Those vector objects manually constructing shapes,
passed wrong pointer as the cacahing key instead of root node pointer.
This fixes it.
Summary:
Coverity reports that we potentially leak char *s here. If we do not
have 'data_ret', then the malloc'd 's' sould be freed as we are not
going to use it.
Fixes Coverity CID1396949
@fix
Reviewers: raster, cedric, bu5hm4n, zmike
Reviewed By: bu5hm4n
Subscribers: #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8523
- don't generate and re-link on every make rule
- fix distcheck
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Reviewed-by: YeongJong Lee <yj34.lee@samsung.com>
Differential Revision: https://phab.enlightenment.org/D8506
Summary:
log level=2 is the warning level, which is not super useful since
there's currently billions of eo warnings occuring in every function
call
Reviewers: cedric, segfaultxavi
Reviewed By: cedric, segfaultxavi
Subscribers: segfaultxavi, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8515
Summary:
passing null here causes any log message to crash the app and is probably not
the intended result
Reviewers: cedric, segfaultxavi
Reviewed By: cedric, segfaultxavi
Subscribers: cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8513
Summary:
1- Speed up detecting new lines.
```
if (!strncmp(text, "<br", 3) || !strncmp(text, "<ps", 3))
```
This will cause 6 comparisons (if one of conditions did not meet), or at least 3 comparisons.
this is changed to
```
if (!strncmp(text, "<", 1))
```
2- Speedup detecting lines
If this condition is true, we should increment the string for next iteration 3 times, not just one
```
if (!strncmp(text, "<br", 3) || !strncmp(text, "<ps", 3))
```
if '<' founded then 'pr' or 'br', we will skip 3 characters for next iteration.
```
if (!strncmp(text, "<", 1))
{
text++;
len--;
if (!strncmp(text, "br", 2) || !strncmp(text, "ps", 2))
{
text += 2;
len -= 2;
```
3- Prevent reading invalid memory out of the string
```
if (text[3] == '>' || ((text[3] == '/') && (text[4] == '>')))
```
string could reach last char in string (original string ends with "<br")
but now we will check if remaining string length allow comparison :
```
if (text[0] == '>' || (len > 1 && ((text[0] == '/') && (text[1] == '>'))))
```
Test Plan:
```
static int
oldFunc(const char *text)
{
if (!text)
return 0;
while (*text)
{
if (!strncmp(text, "<br", 3) || !strncmp(text, "<ps", 3))
{
if (text[3] == '>' || ((text[3] == '/') && (text[4] == '>')))
{
return 1;
}
}
text++;
}
return 0;
}
static int
newFunc(const char *text)
{
if (!text)
return 0;
char *pTemp = (char *)text;
while (pTemp = strchr(pTemp, '<'))
{
pTemp++;
if (!strncmp(pTemp, "br", 2) || !strncmp(pTemp, "ps", 2))
{
pTemp += 2;
if (pTemp[0] != '\0' && (pTemp[0] == '>' || (pTemp[0] == '/' && pTemp[1] == '>')))
{
return 1;
}
}
}
return 0;
}
int main()
{
int counter = 1000;
srand(time(NULL));
char pStr[50001] = {0};
char AllChars[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789<>";
int AllCharsLen = strlen(AllChars);
for (int i = 0; i < 50000; i++)
pStr[i] = AllChars[rand() % AllCharsLen];
clock_t start, end;
double total_Time1 = 0;
int i;
for (int j = 0; j < 3; j++)
{
if (j == 0)
{
printf("random String\n");
}
else if (j == 1)
{
printf("With Random <br/>\n");
int location = rand()%(5000 - 5);
pStr[location++] = '<';
pStr[location++] = 'b';
pStr[location++] = 'r';
pStr[location++] = '/';
pStr[location++] = '>';
}
else if (j == 2)
{
printf("With Random <ps>\n");
int location = rand()%(5000 - 4);
pStr[location++] = '<';
pStr[location++] = 'p';
pStr[location++] = 's';
pStr[location++] = '>';
}
start = clock();
for (i = 0; i < counter; i++)
oldFunc(pStr);
end = clock();
total_Time1 = ((double)(end - start)) / CLOCKS_PER_SEC;
printf("original = %f has new Line = %i\n", total_Time1, oldFunc(pStr));
start = clock();
for (i = 0; i < counter; i++)
newFunc(pStr);
end = clock();
total_Time1 = ((double)(end - start)) / CLOCKS_PER_SEC;
printf("modified = %f has new line = %i\n\n", total_Time1, newFunc(pStr));
}
}
```
output:
random String
original = 2.523000 has new Line = 0
modified = 0.090000 has new line = 0
With Random <br/>
original = 0.081000 has new Line = 1
modified = 0.003000 has new line = 1
With Random <ps>
original = 0.016000 has new Line = 1
modified = 0.001000 has new line = 1
Reviewers: zmike, woohyun, bowonryu
Reviewed By: zmike
Subscribers: bu5hm4n, segfaultxavi, zmike, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D8497
this resolves a lot of cases where focus_highlight API was called on a
object, which is not a efl_ui_win object. With this patch we ensure that
the object is always a window.
Reviewed-by: Xavi Artigas <xavierartigas@yahoo.es>
Differential Revision: https://phab.enlightenment.org/D8476
the problem with the previous implementation (just redirect the calls to
the widget_parent then to the efl_parent is that after invalidate its
impossible to find the window where the widget is in. However, there are
cases where we want to have access to the window of the widget, for
example, to invalidate focus highlight etc..
The window of a widget is always constant, and cannot be changed (as the
evas object cannot hop accross different evas)
Reviewed-by: Jaehyun Cho <jae_hyun.cho@samsung.com>
Differential Revision: https://phab.enlightenment.org/D8475
If docs are present at property and set/get levels only one is used.
Ref T7717
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D8494
exit_on_close and exit_on_all_windows_closed deserve a bit of clarification
since they have very similar meanings.
Also, add proper Eina.Value doc references.
Ref T7717
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D8511
Eina.Values are built-in eolian types, accessed through any_value and
any_value_ptr. However, these types cannot be used in doc references.
Adding a placeholder extern struct Eina.Value causes no harm, and will allow
referencing the type from EO docs later on.
In C#, Eina.Value is defined in the manual binding code so the doc reference
resolves to a valid type.
Reviewed-by: Marcel Hollerbach <mail@marcel-hollerbach.de>
Differential Revision: https://phab.enlightenment.org/D8510
before the refactoring of the disabled property, there was no way to
enable a widget which has a disabled tree. This here however enables
this to work again like this. The user will be told with an error
message. The integraty of the property is maintained accross reparents.
Reviewed-by: Jaehyun Cho <jae_hyun.cho@samsung.com>
Differential Revision: https://phab.enlightenment.org/D8459