00:18:37 GMT <obZen> Can anyone help me with destination to Redis using syslog-ng?
09:10:43 GMT <deneme> hello guys
12:03:20 GMT <King_DuckZ> I'm testing my c++ library, on my system I can set 1.5 mil items in 5.2 seconds... I need to compare that to some other data now
12:04:39 GMT <King_DuckZ> does it sound like it's any good at all to anybody?
12:17:31 GMT <minus> King_DuckZ: compare against redis-benchmark maybe?
12:17:55 GMT <King_DuckZ> minus: yeah, I just tried this redis-benchmark -s /tmp/redis.sock -d 19 --dbnum 0 -c 1 -n 1500000 -P 1000 -t set -r 100000
12:18:18 GMT <King_DuckZ> I *think* it's doing the same thing as my code, but it only takes 1.2 sec
12:18:35 GMT <King_DuckZ> so either it's doing something different or I have some huge bottleneck in my library
12:20:22 GMT <King_DuckZ> 1.5 sec if I do -r 1500000
12:41:08 GMT <King_DuckZ> something's not right... I do flushdb, so dbsize=0, then I run "redis-benchmark -s /tmp/redis.sock -d 9 --dbnum 0 -c 1 -n 10 -P 1000 -t set -r 10000000"
12:42:06 GMT <King_DuckZ> I've got -n 10 in there, so I expect dbsize to be 10 (or less), but it's always 1000... so either I came up with the wrong parameters or... what's going on?
12:42:55 GMT <Habbie> show us please
12:44:11 GMT <King_DuckZ> show what?
12:44:29 GMT <Habbie> your dbsize
12:44:35 GMT <Habbie> although, by now i am also seeing it on my system
12:44:39 GMT <King_DuckZ> 1000
12:45:37 GMT <Habbie> it's the -P 1000
12:45:50 GMT <King_DuckZ> http://alarmpi.no-ip.org/haste/suhop.txt
12:47:49 GMT <King_DuckZ> yeah, so reading at "Using pipelining" here https://redis.io/topics/benchmarks the meaning I get is that the client will never send the 1001th request until any of the first 1000 returned from the server
12:47:53 GMT <King_DuckZ> is that wrong?
16:16:26 GMT <King_DuckZ> I'm reviewing my code, and looking at here https://github.com/KingDuckZ/incredis/blob/master/src/batch.cpp#L172 I'm not sure it's optimal
16:17:25 GMT <King_DuckZ> elsewhere in my code I create a new thread and assign it to libev, so it ends up doing IO in hiredis, if I got it right
16:18:34 GMT <King_DuckZ> that line from my code is directly behind clients' call to the run() method, and I want that to be non-blocking
16:20:04 GMT <King_DuckZ> if hiredis_run_callback for some reasons has to wait for a bit, the wait would be done in the IO thread, is that right?
16:29:35 GMT <minus> have you considered using asio instead of libev?
17:14:11 GMT <King_DuckZ> minus: boost asio?
17:15:31 GMT <minus> "yes". you can also use asio without boost, it's header-only too
17:16:23 GMT <King_DuckZ> I didn't know about asio back when I wrote my library... what's the benefit compared to libev?
17:17:51 GMT <minus> it's a C++ lib; might be easier to work with than libev's C interface
17:18:00 GMT <minus> you can drop lambdas into it for example
17:19:18 GMT <King_DuckZ> it wasn't easy to get libev working right, that's for sure :p
17:19:48 GMT <King_DuckZ> but it's working now, I don't want to touch at that part unless I have to
23:38:04 GMT <slester> Is there a way of saying, "if this isn't set already, set it to x. If it is set, do a BITOP AND with this value"