| 02:27.45 | *** join/#asterisk-dev infobot (ibot@208.53.50.136) |
| 02:27.46 | *** topic/#asterisk-dev is Asterisk Development Discussion -=- http://www.asterisk.org/developers -=- Tier 2 and 3.14159265 support is in #asterisk -=- Check out our blog! blogs.asterisk.org -=- Follow on Twitter at @AsteriskDev |
| 03:56.37 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 04:23.33 | *** join/#asterisk-dev jkroon (~jkroon@165.16.203.126) |
| 05:16.00 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 05:54.35 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 05:55.46 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 06:21.18 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 06:30.09 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 06:37.48 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 06:57.46 | *** join/#asterisk-dev coreyfarrell (~coreyfarr@24-177-250-191.dhcp.nwtn.ct.charter.com) |
| 06:57.46 | *** mode/#asterisk-dev [+o coreyfarrell] by ChanServ |
| 07:11.04 | *** join/#asterisk-dev drc (~drc@unaffiliated/drc) |
| 07:22.16 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 07:23.44 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 07:37.51 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 08:03.06 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 08:07.56 | *** join/#asterisk-dev hehol (~hehol@gatekeeper.loca.net) |
| 08:35.11 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 09:07.53 | *** join/#asterisk-dev CELYA_ (~Thunderbi@2a01:e35:2f15:8190:c587:2d11:df4b:2dde) |
| 09:57.19 | *** join/#asterisk-dev Gugge (gugge@92.246.2.105) |
| 09:59.19 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 10:55.16 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 10:58.29 | file | coreyfarrell: I'll add it to my list to eventually do - I'm taking it easy today after a mess yesterday, may actually poke at code :P |
| 11:00.46 | coreyfarrell | file: you mean confirming the sip transfer issue exists with older pjsua or that all tests will pass with asterisk compiled against pjproject-2.8 but tested with pjsua-2.7.2? |
| 11:00.52 | file | yes |
| 11:01.19 | file | or to be more specific, rolling back Asterisk + pjsua to an older version and testing |
| 11:03.27 | coreyfarrell | yesterday I ran the full testsuite against Asterisk 13/HEAD but with the two pjproject-2.8 patches reverted. had a crash in libspandsp for one of the fax tests, 2 other seeming unrelated failures (that I've never seen in jenkins). everything else passed. |
| 11:03.49 | coreyfarrell | unfortunately I'm already running a different test so, I don't have the logs from yesterdays run |
| 11:04.07 | coreyfarrell | but it definitely went better with pjproject 2.7.2 |
| 11:05.05 | file | with 2.8 pjsua and 2.7.2 Asterisk, or 2.7.2 for both? |
| 11:06.25 | coreyfarrell | my CI container does not install pjproject so 2.7.2 for both (pjsua is from --with-pjproject-bundled) |
| 11:06.35 | file | ack |
| 11:06.44 | file | yeah, I think it's just test level failures |
| 11:07.28 | coreyfarrell | I think definitely worth seeing what happens with pjsua-2.7.2 but Asterisk using bundled 2.8 |
| 11:08.34 | coreyfarrell | might have to force replacement of the bundled pjsua binary, I'm not sure if the testsuite will prefer the bundled pjsua over one installed to the standard system location. |
| 11:09.24 | file | I believe it does |
| 11:34.04 | *** join/#asterisk-dev tsearle (~tsearle@67.231.202.2) |
| 11:34.05 | *** mode/#asterisk-dev [+o tsearle] by ChanServ |
| 11:49.36 | coreyfarrell | do we have an easy way to schedule a callback to happen on fully booted? |
| 11:50.42 | file | are you asking from Asterisk, or from a test |
| 11:53.34 | coreyfarrell | inside asterisk.. to defer a warning until startup is complete. |
| 11:53.54 | file | there is nothing explicitly but you can do it by subscribing for the stasis event |
| 11:54.02 | file | it's done in a few PJSIP modules |
| 11:55.14 | coreyfarrell | ah I see thanks |
| 12:05.06 | coreyfarrell | is there an offical migration guide for chan_sip -> chan_pjsip? |
| 12:05.41 | file | for end users? |
| 12:06.20 | file | https://wiki.asterisk.org/wiki/display/AST/Migrating+from+chan_sip+to+res_pjsip |
| 12:06.26 | file | https://wiki.asterisk.org/wiki/display/AST/Configuring+res_pjsip |
| 12:06.30 | file | also lots'o'stuff linked from there |
| 12:08.29 | *** join/#asterisk-dev CELYA_ (~Thunderbi@2a01:e35:2f15:8190:c587:2d11:df4b:2dde) |
| 12:11.58 | *** join/#asterisk-dev CELYA_ (~Thunderbi@2a01:e35:2f15:8190:c587:2d11:df4b:2dde) |
| 12:29.27 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 12:29.31 | coreyfarrell | waits for seanbright to start yelling at me |
| 13:13.21 | seanbright | hmm? |
| 13:14.27 | seanbright | i'm not really a yeller |
| 13:15.12 | seanbright | for the record: i don't have anything against deprecating chan_sip |
| 13:15.50 | seanbright | i just didn't think having a 30 minute discussion about it in a 2 hour open discussion window was a good use of our time |
| 13:18.27 | seanbright | coreyfarrell: re: your patch |
| 13:19.07 | seanbright | based on last year's discussion, i think it would be better to show a similar warning for any loaded deprecated module |
| 13:19.11 | *** join/#asterisk-dev sduthil (~sduthil@modemcable094.94-70-69.static.videotron.ca) |
| 13:19.19 | seanbright | it might be better to generalize it |
| 13:19.21 | coreyfarrell | seanbright: you give everyone a hard time so I figured I'd return the favor. so really what you're saying is you want us to just get it over with. |
| 13:19.40 | seanbright | coreyfarrell: i just don't want to waste our time talking about it - one way or the other |
| 13:20.29 | seanbright | we get all of the developers from digioma and the community together in a room and we spend too much time talking about something that doesn't matter |
| 13:20.44 | seanbright | </rant> |
| 13:20.53 | coreyfarrell | it'll keep coming up until it's done so the only way to end the discussion is to just make it happen |
| 13:21.00 | seanbright | +1 |
| 13:23.49 | seanbright | do all of the constituent source files even need a MODULEINFO block? |
| 13:25.03 | seanbright | here's what i think we do: |
| 13:25.07 | seanbright | 1) deprecate chan_sip |
| 13:25.37 | seanbright | 2) generalize your deprecation notice to point to a page on the wiki that lists deprecated modules and migration guides |
| 13:25.50 | seanbright | 3) show said notice for all loaded deprecated modules |
| 13:26.15 | seanbright | 4) POTENTIALLY show that message everytime you connect with asterisk -r |
| 13:26.38 | seanbright | what do you think? |
| 13:27.16 | seanbright | if you put it in loader.c, you can overload ast_process_pending_reloads() to print the notices |
| 13:27.26 | file | I think on 4 there'd need to be an option to disable such behavior or else it'll be a nuisance for people who understand and acknowledge the fact |
| 13:27.34 | seanbright | that's fair |
| 13:27.53 | seanbright | or maybe we only show it the first time you asterisk -r |
| 13:30.40 | coreyfarrell | if we do a wiki that lists deprecated modules I think it should be a different link per major branch. |
| 13:31.30 | seanbright | or it could just link to a page that lists the major branches so you have to drill down |
| 13:31.46 | seanbright | i think that's a fairly minor issue regardless |
| 13:32.15 | coreyfarrell | I don't think I like the #4 idea but an option to disable it would be good enough I think. obviously as long as it doesn't show for 'asterisk -rx' |
| 13:32.43 | seanbright | it needs to be "in your face" to some degree |
| 13:33.08 | seanbright | i don't look at logs unless something goes wrong |
| 13:33.18 | file | people ignore lots of what is right in front of them |
| 13:33.37 | seanbright | ok |
| 13:33.48 | seanbright | i've given my opinion |
| 13:33.54 | coreyfarrell | fair point.. so yeah an option to disable it would be fine. I guess the next question would we do this in 13/16? |
| 13:34.00 | file | (that's not an opinion of itself, just an observation from my interactions everywhere) |
| 13:34.20 | file | I've literally copy/pasted an error message in the community forums for someone and told them to read it, because it said the precise reason |
| 13:34.28 | seanbright | re: 13 & 16, i don't think so |
| 13:34.40 | *** join/#asterisk-dev fblackburn (~fblackbur@modemcable094.94-70-69.static.videotron.ca) |
| 13:34.44 | seanbright | WELL |
| 13:35.00 | seanbright | no, i think we only do it in 17+ |
| 13:35.08 | seanbright | again, my opinion |
| 13:38.35 | *** join/#asterisk-dev cresl1n (uid299068@asterisk/libpri-and-libss7-expert/Cresl1n) |
| 13:38.35 | *** mode/#asterisk-dev [+o cresl1n] by ChanServ |
| 13:47.28 | *** join/#asterisk-dev pchero (~pchero@87.213.247.82) |
| 13:56.51 | *** join/#asterisk-dev Tim_Toady (~fuzzy@snf-33276.vm.okeanos.grnet.gr) |
| 14:05.47 | *** join/#asterisk-dev qxork (~qxork@unaffiliated/qxork) |
| 14:20.46 | *** part/#asterisk-dev qxork (~qxork@unaffiliated/qxork) |
| 14:27.14 | *** join/#asterisk-dev kharwell (~kharwell@67.231.202.2) |
| 14:27.15 | *** mode/#asterisk-dev [+o kharwell] by ChanServ |
| 14:48.55 | *** join/#asterisk-dev rmudgett (rmudgett@nat/digium/x-lebofjvchqjguylf) |
| 14:48.55 | *** mode/#asterisk-dev [+o rmudgett] by ChanServ |
| 15:14.34 | *** join/#asterisk-dev snuff-work (~snuff-wor@202-161-112-134.tpgi.com.au) |
| 15:14.34 | *** mode/#asterisk-dev [+o snuff-work] by ChanServ |
| 15:47.01 | *** join/#asterisk-dev sahmed (~sahmed@67.231.202.2) |
| 17:06.02 | *** join/#asterisk-dev tsearle (~tsearle@67.231.202.3) |
| 17:06.02 | *** mode/#asterisk-dev [+o tsearle] by ChanServ |
| 17:22.12 | *** join/#asterisk-dev dandann00dle (uid120491@gateway/web/irccloud.com/x-skwupulybogwpjuz) |
| 17:56.24 | seanbright | rmudgett: around? |
| 17:56.26 | *** join/#asterisk-dev libardi (~libardi@177.194.124.58) |
| 17:56.45 | rmudgett | I am triangle |
| 17:58.26 | seanbright | based on the very limited research i have done, it seems that we need to use the atomic intrincis both when writing and when reading values |
| 17:59.00 | seanbright | i don't know enough about the memory model to know if that is true or not |
| 17:59.22 | seanbright | but this: https://stackoverflow.com/a/24149439/21926 |
| 17:59.31 | seanbright | seems to indicate that we should be using them in both cases |
| 17:59.43 | rmudgett | It depends what you are reading it for if a partial write would be a problem. |
| 18:04.12 | seanbright | i guess i am going for 'correctness' |
| 18:05.00 | seanbright | and by 'correctness' i mean: a partial write would be a problem |
| 18:12.59 | rmudgett | besides any read you do is going to be stale. After you read it it could be updated before you have a chance to use the value. |
| 18:13.51 | seanbright | well, stale is one thing |
| 18:14.37 | seanbright | i want to read the correct value of a variable that is written using the atomic intrinsics |
| 18:15.27 | seanbright | i know what you are saying though |
| 18:15.40 | seanbright | the fetchadd intrinsics eliminate a CAS |
| 18:18.10 | seanbright | anyway. good times. |
| 18:51.59 | *** join/#asterisk-dev tsearle (~tsearle@67.231.202.2) |
| 18:52.00 | *** mode/#asterisk-dev [+o tsearle] by ChanServ |
| 18:57.09 | *** join/#asterisk-dev kharwell (~kharwell@67.231.202.4) |
| 18:57.09 | *** mode/#asterisk-dev [+o kharwell] by ChanServ |
| 19:10.57 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 19:20.02 | *** join/#asterisk-dev kharwell (~kharwell@67.231.202.4) |
| 19:20.02 | *** mode/#asterisk-dev [+o kharwell] by ChanServ |
| 19:20.56 | coreyfarrell | seanbright: atomics can prevent reordering of code (maybe other optimizations as well). https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html |
| 19:21.25 | seanbright | so in real life i write java code |
| 19:21.36 | seanbright | there are a lot of synchronization methods in java |
| 19:22.05 | seanbright | but the 'rules' are such that you need to use them both when reading and with writing |
| 19:22.21 | coreyfarrell | __atomic_store and __atomic_load exist but no __sync variant of those existed. I'm not entirely clear the benefit of using __atomic_load other than if you require strict code ordering |
| 19:22.22 | seanbright | in order to guarantee non-dirty reads |
| 19:22.50 | seanbright | with the __sync family you can do a fetchadd of 0 |
| 19:23.29 | *** part/#asterisk-dev kharwell (~kharwell@67.231.202.4) |
| 19:23.34 | coreyfarrell | possible that the "must use for read and write" is for cross-platform compatibility. not all platforms support hardware atomics. |
| 19:23.43 | seanbright | right |
| 19:24.15 | seanbright | i think the answer to my question is: you should use atomics for reading and writing to guarantee 'correctness' |
| 19:24.24 | seanbright | that's all i was trying to determine |
| 19:28.16 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 19:31.42 | coreyfarrell | seanbright: are you trying to do something specific with atomics in asterisk? reduced channel locking? :) |
| 19:34.03 | seanbright | no |
| 19:34.18 | seanbright | it's not an asterisk specific question |
| 19:34.57 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 19:36.06 | seanbright | i think reducing channel locking is pretty much impossible |
| 19:36.20 | seanbright | and i don't have the expertise |
| 19:40.29 | coreyfarrell | I wouldn't want to try without targeting a specific performance critical code path that is being slowed by locking. |
| 19:40.31 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 19:42.46 | seanbright | agreed. |
| 19:42.57 | coreyfarrell | rmudgett: not sure if you're doing any reviews this week but I'd appreciate your thoughts on https://gerrit.asterisk.org/10447 (also let me know if you think it should go into 16). |
| 19:43.28 | coreyfarrell | unless I'm mixed up I think that patch is at your request (or maybe it was file) |
| 19:44.52 | rmudgett | coreyfarrell: That patch was an idea I had. |
| 19:46.10 | seanbright | way to take credit for coreyfarrell's work |
| 19:46.11 | seanbright | heh |
| 19:46.33 | coreyfarrell | I did the work but it definitely wasn't my idea |
| 19:49.02 | seanbright | coreyfarrell: if you are interested in performance, intel has free tools for OSS contributors |
| 19:49.05 | seanbright | https://software.intel.com/en-us/qualify-for-free-software/opensourcecontributor |
| 19:55.44 | coreyfarrell | not sure I qualify for the license (legally speaking). https://registrationcenter.intel.com/en/forms/?programid=OPNSRC&productid=2908 with the "not paid and/or compensated by anyone". |
| 19:56.21 | seanbright | you're paid for your work on asterisk? |
| 19:56.26 | seanbright | must be NICE |
| 19:56.28 | seanbright | heh |
| 20:00.20 | coreyfarrell | not for everything I do but for some of it. I'm paid to ensure stable operation of an Asterisk based solution. |
| 20:00.43 | seanbright | gotcha |
| 20:06.43 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 20:09.56 | seanbright | https://github.com/asterisk/asterisk/commit/298d015828a638a8ab0f07f8f8205ef04026bdd8#diff-40a604bec35f4f3f242bf94de39996c5 |
| 20:10.11 | seanbright | can someone tell me what exactly that does other than set a 'try_include' variable to 1? |
| 20:11.13 | file | uses it at the bottom to make something non-fatal I think? |
| 20:11.52 | seanbright | oh, right |
| 20:12.00 | seanbright | yikes |
| 20:12.08 | seanbright | THANKS |
| 20:18.21 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 20:24.52 | coreyfarrell | tryinclude is pretty nice, I use it in the configs I install with my Asterisk RPM like '#tryinclude sip.d/*.conf' so the other program which generates accounts can just write sip.d/agents.conf without ever touching sip.conf. |
| 20:59.14 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 21:03.36 | *** join/#asterisk-dev fblackburn (~fblackbur@qubcpq0957w-lp140-01-76-64-186-214.dsl.bell.ca) |
| 21:28.27 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |
| 22:19.55 | *** join/#asterisk-dev fblackburn (~fblackbur@qubcpq0957w-lp140-01-76-64-186-214.dsl.bell.ca) |
| 22:26.07 | *** join/#asterisk-dev fblackburn (~fblackbur@qubcpq0957w-lp140-01-76-64-186-214.dsl.bell.ca) |
| 23:50.02 | *** join/#asterisk-dev awkwardpenguin (~awkwardpe@cpe-172-118-246-147.socal.res.rr.com) |