Summary:
In the case when you have multiple future in flight related to one object, you
couldn't use the previous version of efl_future_then. Now all function calls
take a void* pointer that allow multiple future to have their private data
request data accessible in all the callback.
This should not break released API as Eo.h is not released yet and so
was efl_future_Eina_FutureXXX_then.
Depends on D7332
Reviewers: felipealmeida, segfaultxavi, vitor.sousa, SanghyeonLee, bu5hm4n
Reviewed By: segfaultxavi
Subscribers: #reviewers, #committers
Tags: #efl
Maniphest Tasks: T7472
Differential Revision: https://phab.enlightenment.org/D7379
Summary:
if the user or system attempts to cancel this thread then it should
stop blocking and exit in order to avoid potentially exiting after
efl has expected ecore-con to stop being active
@fix
fix T7041
Depends on D6354
Reviewers: ManMower, devilhorns
Reviewed By: ManMower
Subscribers: cedric, #committers
Tags: #efl
Maniphest Tasks: T7041
Differential Revision: https://phab.enlightenment.org/D6355
This changes a lot of things all across the EFL. Previously,
methods tagged @const had both their external prototype and
internal impl generated with const on object, while property
getters only had const on the external API. This is now changed
and it all has const everywhere.
Ref T6859.
When setting EOS on the dialoer, it's possible that it will also get
automatically closed (by a callback). This results in safety check error
messages, while everything is fine (at least I believe it is).
Gcc complains that 'flags' here may be used uninitialized. In looking
at the code, 'flags' does not seem to be needed in the debug prints
here. If we keep and initialize the variable to 0 during declaration,
it would only ever print out 0 anyway as 'flags' is never changed in the
code.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
so ecore_con/efl_net were using the standard ecore_thread thread pool
for doing things like dns lookups (that can take multiple minutes
until timeouts) and actual http transactions. similarly they can block
thread workers for long periods or indefinitely thus basically
blocking the whole eocre_thread pool and stopping others from sharing
it. the best solution we have right now is to bypass the thread pool
queue and have dedicated threads for these actions. what we should
have is a dedicated thread pool with each thread taking on N
connections (via select etc.) and the ability to create and destroy
thread pools for specific tasks so you can separate the work out from
other work. but that is basically a redesign of our thread pool infra
so let's do the quick solution here until that day comes.
this partially addresses D4640
a dedicated thread per image load though is going to be a lot nastier...
After we call curl_multi_socket_action() we must call
_efl_net_dialer_http_curlm_check() in order to call
curl_multi_info_read() and be notified of handles that were
finished. Otherwise the handle is gone and we'll be waiting for an
action that will never happen.
Fixes T5079
so here's the ugly problem. libproxy. yes. we've discussed memory
usage (e.g. it may have to execute javascript and pull in lots of deps
etc.) but we dlopene'd on the fly. ok... but this didn't solve another
issue i hit:
libproxy was causing enlightenment to abort(). some internal bit of
libproxy was raising a c++ exception. this wasn't caught. this causes
an abort(). takes down your entire desktop. FANTASTIC. this is bad. i
wouldnt' expect a library we depend on to be THIS anti-social but
libproxy seemingly is. it SHOULd catch its error sand just propagate
back to us so we can handle gracefully.
there reall is no way around this - isolate libproxy. it's even worse
that libproxy can load arbitrary modules that come from anywhere sho
who knows what issues this can cause. isolation is the best solution i
can think of.
so this makes an elf+net_proxy_helper we spawn the first time we need
a proxy lookup. we re-use that binary again and again until it exits
(it should exit after 10 seconds of being idle with no requests coming
in/pending). it'll respawn again later if needed. this involves now
the efl net threads having to marshall back to mainloop to do the
spawn and to write to the proxy process (reading is done by async exe
data events and the data is passed down a thread queue to the waitng
efl net thread). if the exe dies with pending requests unanswered then
it's respawned again and the req's are re-sent to it... just in case.
it has a limit on how often it'll respawn quickly.
this seems to work in my limited testing. this ALSO now isolates
memory usage of libproxy to another slave process AND this process
will die taking its memory with it once it's been idle for long
enough. that;s also another good solution to keeping libproxy impact
at bay.
On destructor we're not supposed to emit events, I even thought that
would be implicit, but it's not. If we do, for example an event
handler that would 'efl_del()' on "EFL_IO_CLOSER_EVENT_CLOSED" would
trigger too-many unrefs.
We do not need to keep a "only_head" flag, but we must set
CURLOPT_NOBODY instead of going the "CUSTOMREQUEST" route, otherwise
curl won't follow redirects, etc.
With the last patch to fix delete-from-curl callback it went too much,
considering it was always dead (in the test scenario it was, so it was
"right"), but broke normal cases.
This was annoying to identify as the sequence is kinda difficult to
get, but Terminology was doing a HEAD request and it was triggering
this case in particular.
Fixes T4975.
Windows time_t is not a long, but long-long, then stick with int64_t
so it works everywhere (converts to time_t internally).
And there is no gmtime_r(), then use the gmtime() if not detected.
CURL needs some special curl_easy_setopt() calls to enable automatic
gzip deflate (CURLOPT_ENCODING) and
If-Modified-Since/If-Unmodified-Since logic.
As If-Modified-Since/If-Unmodified-Since requires a timestamp string,
let's expose class methods to handle those.
On Windows SOCKET is unsigned, thus will cause sign errors when
formatting with "%d" or comparing with signed values.
On UNIX it was quiet and easy to miss, thus a new #define can be used
to check for those. It will use 'unsigned long' as SOCKET, thus will
complain out loud and not even work correctly when using pointers on
64bits UNIX on mistakes -- which should improve the situation.
This helped to fix lots of missing conversions, all fixed.
This partially addresses D4357.
for short downloads the CURL handle will be done before the client had
time to read it, like done by efl_io_copier's job. We need to wait it
drain before we emit eos/closed.
Thanks to vtorri for poiting out about close() is not the correct
socket function, we should use closesocket() instead.
Also defined SOCKET to int on Linux so we can use the same 'type' and
avoid lots of ifdef in our code. On Windows it's unsigned, thus would
cause some warnings about incorrect signed comparison.
Defined INVALID_SOCKET=-1 and SOCKET_ERROR=-1 on non-Windows platforms
so we can keep the same construct 'function() == error' and it should
work on POSIX and windows.
I cannot test these on Windows, but the situation should be improved
with this commit.
Efl_Future actually work with weak reference. So you do not need to
set things to NULL, but you actually need to register the memory location
of the future with efl_future_use.
It has been discussed on the ML (thread: "[RFC] rename efl_self") and
IRC, and has been decided we should rename it to this in order to avoid
confusion with the already established meaning of self which is very
similar to what we were using it for, but didn't have complete overlap.
Kudos to Marcel Hollerbach for initiating the discussion and
fighting for it until he convinced a significant mass. :)
This commit breaks API, and depending on compiler potentially ABI.
@feature
When CURLOPT_WRITEFUNCTION returns less then the requested amount,
CURL will fail, not call us back with the remaining data.
Then in such cases we must pause CURL and read nothing.
When unpausing we need to kick curl with timeout action so FD handlers
will be re-arranged.
Last but not least, sync our buffer limit with CURL, otherwise it may
always fail if we're smaller than CURL.
CURL doesn't play nice if handles are deleted or modified while it's
dispatching the callbacks, then we must not touch the CURL* easy
handle in those cases, just dissociate the handle from object and
schedule a job to do the deletion later.
Also, since from CURL callbacks we do not have the reference to the
object, if they are deleted from inside the callback, users of 'pd'
will crash. Thus keep an extra reference while the object and its
private data are in use.
The curl_multi_info_read() is used to notify of errors and
end-of-stream, if we do callback directly from there, the user may
efl_del(dialer), which will result in the "pd->easy" being destroyed
with curl_easy_cleanup() then "cm" and "cm->multi" being destroyed.
Thus postpone that action and keep a list of finished objects, calling
their event handlers which can delete the object (or siblings), thus
ref before dispatching and unref afterwards, taking care to monitor
EFL_EVENT_DEL so we do not use stale objects.
Cedric BAIL
Mike Blumenkrantz
Daniel Kolesa
Carsten Haitzler
Jean-Philippe Andre
Andy Williams
Guilherme Iscaro
Chris Michael
Gustavo Sverzut Barbieri
Tom Hacohen
Vitor Sousa