00:00:33  * dvvquit (Read error: Operation timed out)
00:02:08  * dvvjoined
00:55:10  * indexzerojoined
01:25:34  * indexzeroquit (Quit: indexzero)
01:29:30  * jakejsjoined
05:48:45  * janingequit (Remote host closed the connection)
05:48:55  * janingejoined
05:55:55  * philips_quit (Excess Flood)
05:57:12  * philips_joined
05:59:43  * janinge_joined
06:03:21  * janingequit (Ping timeout: 245 seconds)
06:03:21  * janinge_changed nick to janinge
06:33:25  * coolaj86quit (Read error: Connection reset by peer)
06:33:42  * coolaj86joined
07:25:46  * janingequit (Quit: janinge)
07:43:18  * coolaj86quit (Ping timeout: 264 seconds)
08:01:16  * dnyyquit (Remote host closed the connection)
08:10:28  * dnyyjoined
11:41:06  * mmaleckichanged nick to mmalecki[out]
15:30:12  * janingejoined
15:33:17  <creationix>dvv, is https://github.com/wg/wrk the tool you used to benchmark?
15:33:54  <dvv>yes
15:34:40  <dvv>it provides a solid load on testing server -- ab can't do that and breaks itself
15:36:46  <creationix>yeah, I've been using it
15:36:49  <creationix>simple and fast
15:38:18  <creationix>what;s UV_ECONNRESET?
15:38:36  <dvv>it's when peer goes down suddenly
15:39:00  <creationix>ok, so like a client looses their internet
15:39:01  <dvv>so it's not _the_ error we what to dye of
15:39:08  <dvv>yes
15:40:39  <dvv>i believe it's when the peer did close(fd) on actively sending fd w/o shutdown(fd) first
15:42:07  <creationix>alright, now the server doesn't die when wrk finishes
15:42:12  <dvv>notice also, that the code in question never calls read_stop() -- that way we leak resources
15:42:28  <dvv>it's another chunk of the patch i submitted
15:42:44  <creationix>uv_shutdown doesn't read_stop on it's own?
15:42:55  <dvv>why should it?
15:43:35  <dvv>it tells we finish to write(), not read(), iiuic
15:43:46  <creationix>hmm
15:44:32  <creationix>in that case we shouldn't auto-call read_stop either
15:44:57  <dvv>may be, but we should do so elsewhere
15:44:57  <creationix>I guess the fix is in continuable.lua
15:45:04  <dvv>in luv_close()?
15:45:30  <creationix>these are just bindings to libuv, if libuv doesn't do something, we probably shouldn't either
15:46:09  <creationix>though we do have some abstraction, especially around on_read
15:46:15  <dvv>then it belongs to server code
15:46:56  <dvv>can you clarify why so different logic in tcpstream:close(onclose)?
15:47:06  <dvv>we set a field
15:47:31  <dvv>while in :shutdown(callback) we store callback
15:48:05  <dvv>in short foo:close()() dumps with 'nil value called'
15:48:58  <dvv>you can easily bubble the error by doing `curl -0 -v http://127.0.0.1:8080/` -- notice 0 to force HTTP/1.0
16:22:02  * philips_quit (Excess Flood)
16:22:15  * philips_joined
16:24:54  * coolaj86joined
17:14:58  * kristat__quit (Remote host closed the connection)
17:25:10  * joshthecoderjoined
17:34:43  * TheJHjoined
18:43:36  * janingequit (Quit: janinge)
18:45:22  * janingejoined
18:56:25  * TheJHquit (Ping timeout: 246 seconds)
18:56:42  * TheJHjoined
20:33:05  * mmalecki[out]changed nick to mmalecki
20:53:08  * joshthecoderquit (Quit: Leaving...)
22:36:15  * joshthecoderjoined
22:48:42  * joshthecoderquit (Ping timeout: 264 seconds)
22:51:31  * TheJHquit (Ping timeout: 265 seconds)