CFP are welcomed
But give me a sec
:D
But 1.9.1 by the looks of:
I didn't know travis fires off a new build on tag create...
Are there any limitations to using this: https://github.com/friends-of-reactphp/mysql ?
Since we cannot use PDO in a non blocking environment
@bosunski
@WyriHaximus I get a segfault when I try to test this branch with code coverage on... both on windows and linux...
First core dump on CI was with xdebug and something lead off being something about xdebug and gc collections. After turning it off the test process on the CI still somehow aborts somewhere in the middle without producing a core dump? I don't know, still looking into
Ant (what the CI build agent uses to exec build steps) what Result: 139 at the end of the exec output means...
@WyriHaximus Hmm... :thinking:
Starting program: /usr/bin/php-zts vendor/phpunit/phpunit/phpunit testp.php
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
PHPUnit 7.5.20 by Sebastian Bergmann and contributors.
Random seed: 1579376205
. 1 / 1 (100%)
Time: 77 ms, Memory: 6.00 MB
OK (1 test, 1 assertion)
Program received signal SIGSEGV, Segmentation fault.
destruct_uv_loop_walk_cb (handle=<optimized out>, arg=0x0) at /tmp/pear/temp/uv/php_uv.c:1206
1206 /tmp/pear/temp/uv/php_uv.c: No such file or directory.
(gdb) bt
#0 destruct_uv_loop_walk_cb (handle=<optimized out>, arg=0x0) at /tmp/pear/temp/uv/php_uv.c:1206
#1 0x00007fffec4fa70f in uv_walk () from /usr/lib/x86_64-linux-gnu/libuv.so.1
#2 0x00007fffec72073f in zm_deactivate_uv (type=<optimized out>, module_number=<optimized out>) at /tmp/pear/temp/uv/php_uv.c:2700
#3 0x0000555555b7fc4c in zend_deactivate_modules () at /root/php-src/Zend/zend_API.c:2637
#4 0x0000555555b08062 in php_request_shutdown (dummy=<optimized out>) at /root/php-src/main/main.c:1878
#5 0x0000555555c35350 in do_cli (argc=5, argv=0x555556787090) at /root/php-src/sapi/cli/php_cli.c:1178
#6 0x00005555556e3463 in main (argc=5, argv=0x555556787090) at /root/php-src/sapi/cli/php_cli.c:1404
(gdb)
then its all async
@luiz-brandao-jr Yeah, you want to spawn a new PHP process, that uses a ReactPHP event loop (to still react event-based) but runs blocking code. Since the purpose is to only receive commands, run them and return the response, a blocked event loop isn't a problem here. You only need to keep an eye that everything is there serialized, thus if you need concurrency you need a pool of "workers".
Hi everyone, i am trying to setup a socket server, only for local testing . When using
new Server('0.0.0.0:8009', $loop); and trying to connect with my client (for example vuejs) ws://192.168.0.91:8009/ i always get failed: Error during WebSocket handshake: net::ERR_INVALID_HTTP_RESPONSE - maybe this has something to do with SSL?
i have also added the on listeneres, and can see the client trys to connect:
21.01.2020 13:04:50: new connection 192.168.0.91
21.01.2020 13:04:50: ended 192.168.0.91
21.01.2020 13:04:50: closed 192.168.0.91
so its able to reach the server, but the handshake fails.
So I'm working on a project where I will have around 2k long lasting processes running (think of it this way, I'm creating a web interface for a desktop/cli app). I'm trying to find the best solution given the constraints. The code is stateful, thus the long lasting process. The process will run as long as the user is connected. A process takes a lot of memory and I will always have the risk of having memory leaks. I was thinking of maybe using Threads to reduce memory usage, but there doesn't seem to exist a good solution in PHP for that. What features of ReactPHP could I use for my case?
Some quick estimations suggest 2000 connections * 200 MB RAM per process = 400 GB of RAM to keep all processes running (many variables here, but it's always good to have some rough estimation)
Certainly doable, but you probably don't want to run this on a single host system
If you need help, let's chat: consulting@clue.engineering :-P
@luiz-brandao-jr Neither a one-process-per-user nor a one-thread-per-user is a scalable solution, regardless whether memory is shared or not. What you need is an asynchronous solution and fall back to worker processes for the necessary blocking part, that can't be converted to non-blocking. You probably want to look into different implementations or even languages to suit your needs.