This is heavily inspired from Eo_Id infrastructure. Main change
are that the lower bit are always guaranteed to be zero and ignored
by all function. Also it may be a little bit less efficient in some
case, but we will tune it once we have real life usage of it.
Eo won't be migrated for 1.18 to it as Eo_Id is deeply integrated
and it is quite risky to touch it so close from a freeze. This can
wait.
In case of failure within eina_lock_new() (posix), a
pthread_mutexattr_t would have been left in an initialized state,
without any deinitialization being called.
Consequences would have been implementation defined.
DragonFlyBSD has pthread_setname_np but no pthread_setaffinity_np;
we still need to include pthread_np though.
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Cancelling a promise will fulfill it but won't actually free the memory. This
memory is under custody of the owner, who must either call value_set or
error_set to finish it.
Now when dealing with pointer types, we will not get pointer to
pointer semantics in callbacks and eina_promise_owner_value_set
for Eina_Promise.
It will work as expected:
Eina_Promise_Owner* promise = eina_promise_add();
void* p = malloc(sizeof(T));
eina_promise_owner_value_set(promise, p, &free);
The call to eina_promise_then steals the first ref'count, so it is
possible that the promise is freed after the eina_promise_then,
so we need to eina_promise_ref before eina_promise_then.
We do properly unref promise while calling all the then callback. There
is no need to check it a second time (which actually lead to a 100%
bad access).
T3759
Eina_Error is not passed by pointer anymore, which could cause invalid
pointer access in promise compositions (all and race).
Also added Eina_Promise* to prototypes.
this fixes T3638
@fix
a note... thanks so much to aerodynamik for spotting this. i'm rather
surprised coverity didn't spot this... unless someone said to "shut up
coverity you're wrong" and they should not have.
i also might have expected compilers to spot this too... and add a
warning.
anyway ... this was a seriously subtle bug that could have caused all
kinds of havoc in efl. keys that are different may be compared to be
the same. it could get ordering wrong and sorting thus maybe insert
keys that cannot be found anymore and oh so much more besides.
rthis replaces double a == double b with a macro that keeps a close
enough range using epsilon (which is the error range for a dobule).
this fixes T3245
this should fix T3245
this is basicall where we go double a == double b and due to precision
issues this may not always be right, but this means that the
equivalent now checks for "really close values" rather than perfectly
exact.
@fix
This is done on an attempt to permanently fix our Windows port. Windows
doesn't have MIN/MAX, so we should always do a ifndef/define in every
piece of code that use it. Of course we always forget and it take times
to notice and fix. We have over the year added it in many private
headers, but as the issue continue to raise again and again, I prefer
to get this fixed in our main header.
Added eina_promise_race function that composes multiple
promise objects into a new promise which is fulfilled
when one of the promises are fulfilled, or fails
when one of the promises have failed.
Add a way for users of the promise owner to get notified when a
promise progress is registered. Also added a convenience composition
function that creates a promise which is fulfilled when another
promise has a progress notification.
So it seems we are using Eina_Hash_Iterator quite a lot more than before.
This lead to a huge amount of alloc/free of Eina_Rbtree_Iterator that
was noticable in Enlightenment callgrind trace. This patch make it vanish
from the trace :-)
Fix value_set and error_set signatures which were receiving a
owner. They actually receive the promise and not the owner, this
caused wrong access to memory and were not visible by warnings because
the functions are casted.
This problem caused errors in which it seemed that promise had
actually error'ed when questioned it.
Modify the way hooks are defined and used by promise generation in
Eolian in the Eo API.
Instead of passing macro names as parameters to EO_FUNC_BODY macros,
just re-define the actual hooks when it is needed.
xdg runtime dir is NOT a tmp dir in the normal sense. it's not world
writable nor world readable. only for the user. using
eina_environment_tmp_get() would imply that it is a regular tmp dir,
not a per-user private only runtime dir. that is something else
entirely.
@fix
Add a promise object to allows Eolian interface to include promises
as a way to have asynchronous value return and composibility.
The usage is like this in a .eo file:
class Foo {
methods {
bar {
params {
@inout promise: Promise<int>;
}
}
}
}
Which will create the following API interface:
void foo_bar(Eo* obj, Eina_Promise** promise);
and a Eina_Promise_Owner for the implementation, like this:
void _foo_bar(Eo* obj, Private_Data* pdata, Eina_Promise_Owner* promise);
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
Add a promise object that will allows Eolian interface to include promises
as a way to have asynchronous value return and composibility.
To understand better, let see the coming usage in a .eo file:
class Foo {
methods {
bar {
params {
@inout promise: Promise<int>;
}
}
}
}
Which will create the following API interface:
void foo_bar(Eo* obj, Eina_Promise** promise);
and the equivalent declaration for implementation.
However, the API function will instantiate the Promise for the user
and the implementer of the class automatically. So the user of this
function will treat it as a @out parameter, while the developer of the
function will treat it like a @inout parameter.
So, the user will use this function like this:
Eina_Promise* promise; // No need to instantiate
foo_bar(obj, &promise);
eina_promise_then(promise, callback);
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
Summary:
Use aditional temporary vector for intermedia results in case output vector
the same as target vector in functions:
eina_vector2_transform,
eina_vector2_homogeneous_direction_transform,
eina_vector3_cross_product,
eina_vector3_transform,
eina_vector3_homogeneous_direction_transform
It was in original version (in evas_vecN, module evas_3d_utils.h)
Enrich test suit for this case.
Reviewers: jpeg, cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3795
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
This reverts commit 7f4ea1a79c.
This reverts one of three parts of the try to get sub directory
compilation back into eina. It breaks our distcheck though and I
talked to Cedric about it and he prefers to revert these as we might
need to go another route to bring this functionality back. Details
will come to the mailing list.
This reverts commit 1affc60d00.
This reverts one of three parts of the try to get sub directory
compilation back into eina. It breaks our distcheck though and I
talked to Cedric about it and he prefers to revert these as we might
need to go another route to bring this functionality back. Details
will come to the mailing list.
This reverts commit e26fcbb1dc.
This reverts one of three parts of the try to get sub directory
compilation back into eina. It breaks our distcheck though and I
talked to Cedric about it and he prefers to revert these as we might
need to go another route to bring this functionality back. Details
will come to the mailing list.
This iterator is convenient when you already have a C-Array and you
need to pass this array to a function receiving an Eina_Iterator.
int array[] = {1, 2, 3, 4};
int* array2[] = {&array[0], &array[1], &array[2], &array[3], NULL};
Eina_Iterator* iterator = eina_carray_iterator_new((void**)array);
so the spinlock on the threadqueue block pool it taken on shutdownn,
while the block pool is freed up then its is destroyed, but openbsd
very much doesnt like this and returns an error, so release the lock
before destroying it.
@fix
Before this patch, eina_log would simply record a non-formatted
entry like:
unknown domain -1, original message format 'proxy=%p, obj=%p'
This was not very useful as even if the log domain is invalid, the
message itself might be relevant (often those are ERR logs).
Now the message format is roughly the same as the default format,
except that the line info comes from the original message (and
doesn't refer to eina_log.c).
Backtrace printing will happen at the same level as the original
log level, in order to avoid log pollution in case DBG logs are
printed with an invalid domain (and CRI would trigger bt).
I actually wonder if the logs shouldn't actually be forwarded
to the standard log callback instead of just stderr. This may
be useful for logging with dlog or journald (atm we will simply
lose all logs without a valid domain). This would mean eina_log
itself requires a log domain.
applying this optimization to prevent the same rectangle from being added
or removed repeatedly in succession would result in the rejecting of successive
operations of the same type when the other operation occurred in between.
as an example:
add(0, 0, 100, 100)
del(0, 0, 100, 100)
add(0, 0, 100, 100)
should yield (0, 0, 100, 100), not zero rects and a failure to add the
second rect
this fixes a serious issue in enlightenment where stacking three windows
on top of each other with the first and third windows having the same geometry
would result in the top window receiving no input geometry (oops)
@fix
The prototype means "unspecified arguments". However, the implementation
specifies no parameters (void).
GCC's warning -Wstrict-prototypes complains a lots about this old-style
prototype, spamming the console when compiling more than one compiling
units including Eina.
@fix
If the template is a path, mkstemp and mkdtemp would fail
miserably as they would try to create a file inside
/run/user/1000//path/to/file.XXXXXX even if the path did not
exist.
This patch fixes that by creating temp files inside the sys temp
dir iif the templatename is just a basic name without path
separator.
@fix
We have to use void in a function declaration if we want no function
parameters. Using just empty parenthesis means the function takes an
unspecified number of parameters.
Summary:
Fixes T2949.
In the bug, its been mentioned about updating the documentation, so added this.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric, jpeg
Reviewed By: jpeg
Maniphest Tasks: T2949
Differential Revision: https://phab.enlightenment.org/D3462
The main reason is convenience for debugging when using GDB,
this will give a simple breakpoint for all safety check failures.
Also, this creates a more visible log domain (red).
The EINA_LOG_BACKTRACE thing is aimed at production environments,
so we can extract a backtrace from a log file post-mortem, but not
for continuous development of EFL itself.
I know this should make a few people happy.
Most eina log env vars mean "if loglevel <= val then print log"
but eina_log_backtrace was "if loglevel < val" which I thought
was a bit confusing. The default behaviour is unchanged.
Eina Value Optional can be used to create a eina value that can be set
or be empty and can be embedded in a eina_value_struct.
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Related to T2287.
Log lock errors (printf to avoid eina_log locks) and continue or abort,
conditional on EINA_HAVE_DEBUG_THREADS.
Reviewers: raster, cedric
Subscribers: stefan_schmidt, cedric, seoz
Differential Revision: https://phab.enlightenment.org/D2376
Note - fixed review comments on macro names and some formatting and
error strings too - raster.
Summary:
I have put the common encoding code in the common function. The 2 API's
call them with a flag. Thinking of doing the same way to decode
function also.
T2880
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D3414
This caused functions from eina_util to be undefined at compile time.
The compiler would make implicit casts of return values into ints,
which had major side effects (e.g. segfault edje_cc)
@fix
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
We had this nice shortcuts for multiply and inverse with the identity matrix.
Pity we never used it! The EINA_MATRIX_TYPE_IDENTITY is coming from an enum
without and direct assignments to its internals. Being the first item in the
enum it is most likely will be 0 which makes the whole bitwise AND zero and thus
the optimized path will never get called. If our compiler now decides hew wants
to handle enums differently and does not assign the 0 to the first item this
bitwise operation will be even more screwed. What we really want is to check is
if the type we get for the matrix matches EINA_MATRIX_TYPE_IDENTITY. So better
do this. Made me look into matrix multply and inverse. Fun!
Thanks to smatch for poiting this out.
Required some special treatment to get words (substring) out of an
input Unicode string to a utf8 one. This saves the trouble converting
the whole string: you input an offset in the Eina_Unicode array, and
provide the required length. That's is, now you can extract words and
whatnot in utf8 form.
To save code I wrapped the original one to get the whole length.
Summary:
This API is analogous to c++ substr. It returns a Eina_Strbuf with
substring of passed buf.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3224
This is useful when Eo calls it to get the current frame stack.
This improves the performance of this function by ~50% which meant
from ~3% total CPU time to ~1.5% CPU time (if called at each eo_do).
Now this patch is a bit irrelevant since eo uses __thread instead.
This optimization is still useful for evas_eglGetCurrentXXX.
Summary:
Add new API eina_strftime API in eina_str
@feature
Test Plan: test case and example also updated
Reviewers: tasn, cedric
Reviewed By: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D3148
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Add eina_tmpstr_manage_new, eina_tmpstr_manage_new_length APIs, these APIs create new tmpstr but reuse the input string memory.
@feature
Test Plan: Test case and example updated
Reviewers: tasn, cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3178
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This API would be used in efl_network_websocket.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3200
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This is not an ABI break as we never did provide the broken name symbol, just a typo
in the header that wasn't detected until now.
@fix
Signed-off-by: Vivek Ellur <vivek.ellur@samsung.com>
Reviewers: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D3248
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
This reverts commit 22b45f220c.
eina_file_cleanup always does an eina_tmpstr_del. This is now capable of doing
double or even triple free in some case.
lib/eina/eina_util.c: In function 'eina_environment_tmp_get':
lib/eina/eina_util.c:96:7: warning: 'tmp' may be used uninitialized in this function [-Wmaybe-uninitialized]
if (!tmp) tmp = "/tmp";
^
Summary: If we do not include unistd.h, we end up with implicit
declaration warnings when compiling
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This reverts commit 05c18876b3.
i endabled backtraces by DEFAULT because it is otherwise pretty damned
hard to get them from things like enlightenment, or on tizen apps as
running them on the cmdline and having them work is close to
impossible. yes it's noise. FIX THE ERRORS then. this commit is just
like turning off gcc etc. warnings because the code is too noisy in
compilation with them on. it's a bug. fix it. the bt lets you do that
without gdb or re-executing AGAIN with an env var and HOPING to find
the bug the next time around.
not to mention efl programming docs cover this backtrace and say it
is on by default and how to get useful info out of this. this makes
the elf docs a lie by removing what is documented as default.
@fix
Summary:
@feature
Test Plan:
eina_tmpstr_strftime API can be used to create a temporary string
which is updated with strftime output
eina_tmpstr_steal can be used to get actual string set in eina_tmpstr
Reviewers: cedric
Subscribers: rajeshps, cedric, govi
Differential Revision: https://phab.enlightenment.org/D3048
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
added support to set thread names for debugging. this wraps the gnu
extension to set a thread name of up to 15 chars (16 inc nul byte).
eina_thread_name_set() is the new api.
@feature
in the process i found the autofoo wasnt enabling cpu affinity support
at all in reality, so i had to fix that at the same time.
Summary:
Check for valid rectangle
If any dst or src rectangle has zero width or height,
intersection should not return true.
@fix
Test Plan: Added test cases
Reviewers: cedric, herdsman, Hermet
Reviewed By: Hermet
Subscribers: shilpasingh, cedric
Differential Revision: https://phab.enlightenment.org/D2990