Displaying #traffic-server/2015-12-28.log:

Mon Dec 28 00:02:26 2015  biilmann:Joined the channel
Mon Dec 28 00:08:08 2015  zwoop:Joined the channel
Mon Dec 28 00:12:15 2015  zwoop:PSUdaemon on TS-4009, you feel it's the right thing to fix the plugin, vs the changes made to the "offending" commit ?
Mon Dec 28 00:12:27 2015  zwoop:if it's the former, are we at risk breaking other plugins ?
Mon Dec 28 00:13:04 2015  zwoop:The fix was intentional in that APIs that requires a mutex really ought to set those up
Mon Dec 28 00:16:41 2015  zwoop:my change avoids a global mutex at the expense of plugins having to create mutexes when needed.
Mon Dec 28 00:17:01 2015  zwoop:Maybe we should change
Mon Dec 28 00:17:02 2015  zwoop:+ sdk_assert(((INKContInternal *)_c)->mutex); \
Mon Dec 28 00:17:16 2015  zwoop:to a release assert for now, to let us trip over this nicer.
Mon Dec 28 00:24:17 2015  jpeach:sdk_assert is a release assert
Mon Dec 28 00:26:30 2015  zwoop:oh
Mon Dec 28 00:26:41 2015  zwoop:well, then never mind :)
Mon Dec 28 00:27:01 2015  zwoop:I still think the code is correct, and that we should fix the plugin as the patch proposes
Mon Dec 28 00:27:09 2015  jpeach:agreed
Mon Dec 28 00:49:20 2015  biilmann:Joined the channel
Mon Dec 28 01:27:09 2015  biilmann:Joined the channel
Mon Dec 28 02:18:14 2015  PSUdaemon:zwoop: i commented on the ticket
Mon Dec 28 02:18:20 2015  PSUdaemon:just saw this window
Mon Dec 28 02:18:25 2015  zwoop:ok
Mon Dec 28 02:18:37 2015  PSUdaemon:i just thought before the behavior was to create a mutex if it needed one and you did not provide
Mon Dec 28 02:18:43 2015  PSUdaemon:your patch changed that behavior
Mon Dec 28 02:18:55 2015  PSUdaemon:so i'm fine, just want to be sure we are all on the same page
Mon Dec 28 02:19:00 2015  zwoop:Right, but the documentation always specifies that you must create the mutex
Mon Dec 28 02:19:04 2015  PSUdaemon:fine with the patch as proposed that is
Mon Dec 28 02:19:06 2015  zwoop:for certain types of continuations
Mon Dec 28 02:19:17 2015  PSUdaemon:who reads docs? :-p
Mon Dec 28 02:19:21 2015  zwoop: ;)
Mon Dec 28 02:19:23 2015  zwoop:just saying
Mon Dec 28 02:19:45 2015  zwoop:The problem with the old code is a gigantic global mutex
Mon Dec 28 02:19:53 2015  zwoop:only to make it "easy" to write bad code :)
Mon Dec 28 02:20:06 2015  PSUdaemon:i didn't realize it was global
Mon Dec 28 02:20:12 2015  PSUdaemon:it looks like it made one on the fly
Mon Dec 28 02:20:18 2015  zwoop:it's a global mutex when it has to create it
Mon Dec 28 02:20:19 2015  PSUdaemon:i'm fine with it though
Mon Dec 28 02:20:31 2015  zwoop:it has a global mutex to make sure only one thread creates that mutex
Mon Dec 28 02:23:15 2015  zwoop:we could / should improve the places where it talks about when the mutex is necessary though. It's only a few places where it needs it, and when the continuation uses VIO is one of them if I recall.
Mon Dec 28 02:41:49 2015  zwoop:jpeach TS-4099 you aiming for v6.2.0 right ?
Mon Dec 28 07:07:43 2015  ben1:Joined the channel
Mon Dec 28 07:33:20 2015  mturk:Joined the channel
Mon Dec 28 07:33:21 2015  mturk:Joined the channel
Mon Dec 28 08:01:12 2015  dustywusty:Joined the channel
Mon Dec 28 08:28:58 2015  biilmann:Joined the channel
Mon Dec 28 09:09:07 2015  biilmann:Joined the channel
Mon Dec 28 11:43:35 2015  lev__:Left the channel
Mon Dec 28 13:01:41 2015  niq:Joined the channel
Mon Dec 28 13:01:56 2015  niq:Joined the channel
Mon Dec 28 15:18:15 2015  RamJett:Joined the channel
Mon Dec 28 15:19:19 2015  blattj:Joined the channel
Mon Dec 28 15:21:04 2015  blattj1:Joined the channel
Mon Dec 28 15:38:41 2015  shinrich2:Joined the channel
Mon Dec 28 15:39:34 2015  biilmann:Joined the channel
Mon Dec 28 16:00:37 2015  jumby:Joined the channel
Mon Dec 28 16:40:18 2015  swoc:Joined the channel
Mon Dec 28 16:42:11 2015  yamc:Joined the channel
Mon Dec 28 16:42:59 2015  es:Joined the channel
Mon Dec 28 16:54:56 2015  niq:Joined the channel
Mon Dec 28 17:00:48 2015  dcarlin:Did dual ecdsa/rsa certs ever become a thing?
Mon Dec 28 17:00:53 2015  dcarlin:http://trafficserver-test.readthedocs.org/en/latest/ssl.html#dual-certificate-ecdsa-rsa
Mon Dec 28 17:00:58 2015  dcarlin:just says …
Mon Dec 28 17:01:40 2015  mlibbey:wow, there are a lot of "..." on that page.
Mon Dec 28 17:02:20 2015  dcarlin:I poked around jira for a bit and didn't see anything
Mon Dec 28 17:03:35 2015  dcarlin:https://news.ycombinator.com/item?id=8276801
Mon Dec 28 17:03:49 2015  dcarlin:someone on hacker news says they submitted an ats patch :)
Mon Dec 28 17:46:18 2015  biilmann:Joined the channel
Mon Dec 28 18:45:37 2015  psp:Joined the channel
Mon Dec 28 19:17:26 2015  biilmann:Joined the channel
Mon Dec 28 19:20:47 2015  jpeach:PSUdaemon, zwoop: I'm landing the patch for TS-4009 with minor changes to avoid leaking the mutex
Mon Dec 28 20:02:10 2015  es:Joined the channel
Mon Dec 28 21:26:07 2015  PSUdaemon:jpeach: +1
Mon Dec 28 21:50:39 2015  biilmann:Joined the channel
Mon Dec 28 21:53:05 2015  es:Joined the channel
Mon Dec 28 22:36:34 2015  blattj:Joined the channel
Mon Dec 28 22:38:16 2015  blattj1:Joined the channel

Comments