| 00:17.33 | *** part/#asterisk-dev kharwell (kharwell@nat/digium/x-ofqatoxfnklzhcnw) |
| 01:19.08 | *** join/#asterisk-dev infobot (ibot@208.53.50.136) |
| 01:19.08 | *** 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 |
| 02:58.49 | *** join/#asterisk-dev dandann00dle (~dandann00@c-76-114-64-149.hsd1.tn.comcast.net) |
| 06:37.10 | *** join/#asterisk-dev CELYA_ (~Thunderbi@92.184.116.103) |
| 06:48.05 | *** join/#asterisk-dev jkroon (~jkroon@165.16.203.52) |
| 07:14.47 | *** join/#asterisk-dev tsearle (~tsearle@37.19.10.33) |
| 07:14.47 | *** mode/#asterisk-dev [+o tsearle] by ChanServ |
| 07:23.02 | *** join/#asterisk-dev pchero_work (~pchero@87.213.240.121) |
| 07:24.53 | *** join/#asterisk-dev pchero_work (~pchero@87.213.240.121) |
| 07:54.10 | *** join/#asterisk-dev CELYA_ (~Thunderbi@ip-91-191-153-118.sd-france.net) |
| 09:35.15 | *** join/#asterisk-dev pchero_work (~pchero@87.213.240.121) |
| 10:07.57 | *** join/#asterisk-dev gtjoseph (~gtjoseph@unaffiliated/gtj) |
| 10:07.57 | *** mode/#asterisk-dev [+o gtjoseph] by ChanServ |
| 10:08.15 | *** join/#asterisk-dev pchero_work (~pchero@87.213.240.121) |
| 10:15.33 | *** join/#asterisk-dev pchero_work (~pchero@60.30.148.146.bc.googleusercontent.com) |
| 11:57.56 | *** join/#asterisk-dev pchero_work (~pchero@87.213.240.121) |
| 12:16.49 | *** join/#asterisk-dev ryskov (~jeppelars@213.150.61.162) |
| 12:42.06 | *** join/#asterisk-dev pc-m (~pcm@modemcable094.94-70-69.static.videotron.ca) |
| 12:42.42 | tsearle | ok, so jenkins is failing on a patch that I pushed upstream, apparently on a patch that I also pushed upstream.... but it passes on my local jenkins.... |
| 12:42.57 | gtjoseph | which test? |
| 12:42.59 | tsearle | well, I'm assuming it's my patch breaking things |
| 12:43.44 | tsearle | https://jenkins2.asterisk.org/job/asterisk-gatetests/job/13/237/testReport/junit/channels.pjsip.nat/rewrite_contact/_____pjs2___route_set_reinvite/ |
| 12:44.10 | tsearle | that I pushed upstream because it was just improving code coverage, didn't need a patch to pass |
| 12:44.24 | file | it appears as though the tests that were broken by the patch originally are also broken, or some of them |
| 12:44.42 | tsearle | but that test + the mwi tests are now failing (again) now that my nat patch has been re-gated |
| 12:47.50 | tsearle | SUBSCRIBE as a dialog creating event, can have record route correct? |
| 12:48.40 | file | yes. |
| 12:50.00 | file | interesting, Alexei's change has made shutdown behavior differ and is causing a crash sometimes |
| 12:50.47 | gtjoseph | kharwell will love that |
| 12:52.36 | tsearle | the reuse_connection tests is what I tested the new patch against, and all but one of those got closed, so that's progress |
| 12:52.51 | tsearle | getting hard to find tests that don't require asterisk to be recompiled, heh |
| 12:59.54 | tsearle | can I unfriend jenkins? |
| 13:00.05 | tsearle | or end of life mwi, ;-) |
| 13:00.45 | tsearle | there is a bit of chicken & egg issue with the pjsip_nat code |
| 13:01.02 | tsearle | basically the code to update the contact after dialog has been found doesn't work |
| 13:01.20 | tsearle | so it only works correctly when invoked before the dialog as been resolved |
| 13:01.35 | tsearle | but before you resolve the dialog, you can't tell if there is a saved routeset |
| 13:02.12 | tsearle | if there is a routset, updating in the dialog is sufficent (you don't need to rewrite the message, that part of code can actually be dropped) |
| 13:03.57 | tsearle | and there are other ugly aspects that I've ignored so far with that code |
| 13:04.26 | tsearle | like that it looks at ACK messages too, and ACK can trigger a re-targeting of the contact too |
| 13:06.40 | tsearle | anyhow I guess I'll stop my rant there, but I just feel like there should be a much cleaner way to implement the nat function |
| 13:07.41 | tsearle | as I feel that being invoked 3 times per-request is a bit unnecessary in overhead |
| 13:10.26 | file | if it can be done a better way while maintaining functionality that exists now and not breaking things, then I'm sure we'd be open to it |
| 13:10.39 | *** join/#asterisk-dev coreyfarrell (~coreyfarr@24-181-237-90.dhcp.oxfr.ma.charter.com) |
| 13:10.39 | *** mode/#asterisk-dev [+o coreyfarrell] by ChanServ |
| 13:11.42 | tsearle | indeed, I just don't know all the boxes |
| 13:12.02 | tsearle | my originaly idea was to just only consume when the dialog is found (and update the dialog) |
| 13:12.11 | file | that code is old and has been touched only a few times, so anyone tackling such a thing would need to dig in, explore everything, etc |
| 13:12.43 | tsearle | that's the reason why I expanded it's test coverage before taking this dive, heh |
| 13:13.47 | tsearle | question ref you testsuite |
| 13:13.50 | tsearle | https://jenkins2.asterisk.org/job/asterisk-gatetests/job/13/237/testReport/junit/channels.pjsip.subscriptions.rls.lists.nominal.mwi/batched/_____mwi____basic/ |
| 13:13.59 | tsearle | trying to look for the asterisk logs for this test |
| 13:14.05 | tsearle | but i don't see a dir called mwi |
| 13:14.10 | tsearle | only an "extmwi" |
| 13:14.41 | file | it's in extmwi |
| 13:15.07 | tsearle | ok |
| 13:15.14 | tsearle | thanks :-) |
| 13:15.36 | coreyfarrell | I missed the beginning of this conversation. this isn't about chan_sip? the 'that code is old' comment caught my attention :) |
| 13:15.52 | file | coreyfarrell: no, res_pjsip_nat |
| 13:15.53 | gtjoseph | we have "old" code in pjsip now. :) |
| 13:16.17 | file | coreyfarrell: I do have a fun thing for you... |
| 13:17.02 | coreyfarrell | file: whats up? |
| 13:17.14 | file | https://jenkins2.asterisk.org/job/asterisk-gatetests/job/13/238/artifact/tests/CI/output/extmwi/logs/channels/pjsip/subscriptions/rls/lists/off_nominal/mwi/list_does_not_exist/backtrace_12830.txt |
| 13:17.32 | file | do you remember if we've seen that before? I checked JIRA and didn't come across such a thing |
| 13:20.22 | coreyfarrell | file: I don't recognize that. looks like the unload_module ordering looks bad. should probably clear unsolicited_mwi, do everything else, then unref unsolicited_mwi and set NULL. |
| 13:21.44 | coreyfarrell | oh weird that happens on startup? |
| 13:22.03 | file | it appears to be shutting down |
| 13:22.05 | file | it's odd... |
| 13:23.37 | coreyfarrell | send_initial_notify_all is only called by load_module or in response to FullyBooted (at least in 13) |
| 13:25.35 | coreyfarrell | well it's passed to ast_sip_push_task for run.. so if the system is shutting down too quickly after boot I guess that task could run after the module is unloaded. |
| 13:27.09 | file | meh |
| 13:28.52 | coreyfarrell | file: maybe add ast_module_ref to load_module inside `if (!ast_sip_get_mwi_disable_initial_unsolicited()) {`, add ast_module_unref to the end of send_initial_notify_all? that should prevent the shutdown race. |
| 13:31.54 | file | hrm |
| 13:33.38 | file | I made a JIRA for it |
| 13:49.03 | *** join/#asterisk-dev bford (uid283514@gateway/web/irccloud.com/x-ypwlbmrjgqwtuywp) |
| 13:49.03 | *** mode/#asterisk-dev [+o bford] by ChanServ |
| 13:55.10 | coreyfarrell | file: want to ping your internal bug for the mailing list being down again? |
| 13:56.33 | file | done |
| 14:14.47 | *** join/#asterisk-dev cresl1n (uid299068@asterisk/libpri-and-libss7-expert/Cresl1n) |
| 14:14.47 | *** mode/#asterisk-dev [+o cresl1n] by ChanServ |
| 14:29.14 | *** join/#asterisk-dev dandann00dle (~dandann00@2601:840:8401:d7ee:78e9:77a:c5ce:f3c5) |
| 14:31.01 | *** join/#asterisk-dev pchero_work (~pchero@60.30.148.146.bc.googleusercontent.com) |
| 14:33.10 | pchero_work | Hi, I'm looking for how to set the call-id for outbound call in the Asterisk, may I know where should I look at it? :) Thanks in advance. |
| 14:33.24 | pchero_work | call-id for SIP header. |
| 14:33.42 | pchero_work | for the res_pjsip. |
| 14:33.53 | seanbright | it's not explicitly settable afaik |
| 14:33.57 | file | there is no ability to set it |
| 14:34.16 | pchero_work | Ooops, |
| 14:34.20 | pchero_work | No way? |
| 14:34.24 | seanbright | writes a callid factory interface |
| 14:34.32 | seanbright | no way... yet |
| 14:34.44 | seanbright | until someone writes the code that allows it |
| 14:35.01 | seanbright | i nominate pchero_work |
| 14:35.46 | pchero_work | Hm, is that pjsip's matter? Or Asterisk's thing? |
| 14:35.52 | seanbright | pjsip |
| 14:36.02 | pchero_work | Ah, make sense. |
| 14:36.20 | tsearle | call ids need to be globally cryptographically unique, so usually sip libs don't like to trust other people with such power |
| 14:36.39 | pchero_work | Btw, then for the sending a SIP INVITE request, where should I look at it? I would digging it. :) |
| 14:36.40 | tsearle | expecially as sip libs usually will break pretty badly if the call-id ends up not being unique |
| 14:37.01 | seanbright | https://www.pjsip.org/docs/latest/pjsip/docs/html/group__PJSIP__MSG__HDR.htm#ga48763c2230f3a558b0ae0ab66b368d0a |
| 14:37.13 | tsearle | I would probably look at pjsip_dialog |
| 14:38.06 | pchero_work | @tsearle, Thanks! I'll take a look. :) |
| 14:38.33 | seanbright | pj_create_unique_string |
| 14:38.38 | seanbright | that is what pjsip uses |
| 14:38.56 | file | you have to go through hoops to do it... |
| 14:39.00 | tsearle | https://xkcd.com/221/ |
| 14:39.13 | file | like alter the dialog, remove it from a hash table or two, add it back in, do other stuff |
| 14:39.33 | seanbright | https://trac.pjsip.org/repos/browser/pjproject/trunk/pjsip/src/pjsip/sip_dialog.c#L288 |
| 14:39.37 | tsearle | well I assume if you can control the dialog creation in the first )place |
| 14:39.43 | tsearle | then you don't need to delete/re-add |
| 14:40.00 | file | you don't control it in that level |
| 14:40.32 | tsearle | I'm assuming patching pjsip is an option for him |
| 14:40.49 | pchero_work | Ooops, may it's better to don't. ;) Argggg |
| 14:41.31 | seanbright | would be straight forward i think |
| 14:41.45 | seanbright | not within asterisk per-se, but the pjsip patch |
| 14:42.02 | seanbright | there are only a few places where call-ids are generated |
| 14:42.07 | pchero_work | @seanbright I agree with that. It should be done from pjsip first. |
| 14:42.31 | seanbright | (i apologize for hyphenating per se) |
| 14:43.38 | pchero_work | @seanbright Never mind. I'm just thanks to you. Gave me a lots of instructions. :) |
| 14:43.42 | seanbright | 'cryptographically unique' is not rfc mandated |
| 14:44.00 | seanbright | although globally unique is |
| 14:44.09 | tsearle | ah, sorry for mis-speaking |
| 14:44.17 | seanbright | it's RECOMMENDED |
| 14:44.25 | seanbright | because we need to SCREAM in RFCs |
| 14:44.49 | seanbright | because it's 2018 and the only way to write specifications is in ASCII |
| 14:44.58 | tsearle | never thought of it that way, just always thought of them as programming keywords |
| 14:45.02 | file | seanbright: you MAY scream in RFCs but you MUST use RECOMMENDED or MAY |
| 14:45.15 | file | seanbright: and SHOULD have a tasty lunch |
| 14:45.23 | pchero_work | Right, so, it's not MUST. |
| 14:45.23 | seanbright | i think you covered most of them |
| 14:45.28 | tsearle | actually from what I understood at SipNoc, they write rfc in XML hand have a tool to covert it to ASCII |
| 14:45.37 | seanbright | "Implementations MAY use the |
| 14:45.38 | seanbright | <PROTECTED> |
| 14:45.43 | tsearle | anyhow, ASCII just became an Internet Standard 2 years ago or so |
| 14:45.49 | tsearle | so it's pretty bleeding edge tech |
| 14:51.36 | tsearle | ( January 2015) |
| 14:55.48 | tsearle | and oddly enough it's even newer of a standard than UTF8 (ascii STD-80, utf8 STD-63) |
| 14:56.11 | tsearle | so clearly ASCII deprecates UTF8 and we should all migrate to it :-) |
| 15:02.12 | *** join/#asterisk-dev kharwell (kharwell@nat/digium/x-bubqplggmjjdhqoq) |
| 15:02.12 | *** mode/#asterisk-dev [+o kharwell] by ChanServ |
| 15:45.21 | coreyfarrell | file: FYI I got Bob Cripps response to the code review list. |
| 15:46.00 | file | cresl1n may have let it through |
| 15:51.01 | file | coreyfarrell: I also think keeping all discussion in one place is a good thing >_> |
| 15:51.44 | tsearle | just noticed Bob being mentioned |
| 15:51.50 | coreyfarrell | oh I wasn't disagreeing with that.. just pointing out that the mailinig list might not actually be locked down. |
| 15:52.04 | file | it bounced my response oddly, yet I get posts on it |
| 16:32.38 | cresl1n | I did let it through |
| 16:43.57 | *** join/#asterisk-dev CELYA_ (~Thunderbi@2a01cb0583d31d000459046f3d5f90e1.ipv6.abo.wanadoo.fr) |
| 17:11.20 | *** join/#asterisk-dev AsteriskRoss (259d3426@gateway/web/freenode/ip.37.157.52.38) |
| 17:38.19 | *** join/#asterisk-dev jkroon (~jkroon@165.16.204.34) |
| 19:05.07 | *** join/#asterisk-dev rmudgett (rmudgett@nat/digium/x-eteqteshnhcfheqq) |
| 19:05.07 | *** mode/#asterisk-dev [+o rmudgett] by ChanServ |
| 19:34.19 | seanbright | so i am thinking of patching asterisk to allow bdb to be (optionally) used for astdb instead of sqlite3 |
| 19:34.22 | seanbright | change my mind |
| 19:35.20 | gtjoseph | I have a C implementation of IBM's original ISAM I could let you have. |
| 19:35.25 | seanbright | looking back at the original change[1], it looks like the change to sqlite3 was driven by the bundled bdb having some bugs |
| 19:35.28 | seanbright | [1] https://github.com/asterisk/asterisk/commit/efd040cd11cff99a9dccf54c030a134d9a1e2e07 |
| 19:39.16 | cresl1n | yeah |
| 19:39.19 | cresl1n | I remember this |
| 19:39.23 | cresl1n | file probably does too |
| 19:39.37 | file | hold, I'm transferring a design from my brain to code |
| 19:40.11 | gtjoseph | uhm before we look at replacing the implementation why don't we look at the gawd-awful way we use sqlite3. |
| 19:40.44 | file | the only other way you could use it would be to open multiple instances, if I recall |
| 19:41.08 | gtjoseph | TLS |
| 19:41.31 | file | the actual impact of that and the performance repercussions, I know not |
| 19:42.05 | gtjoseph | yep which is why we should take a real look |
| 19:49.40 | seanbright | i'm not talking about replacing |
| 19:50.03 | seanbright | you should choose at runtime. |
| 19:50.14 | seanbright | we don't need ACID compliance |
| 19:50.18 | seanbright | we just need a fast key/value store |
| 19:51.08 | seanbright | and the "god awful" way we use it is the hacky keep-a-transaction-open-for-long-periods-of-time-and-commit-it-periodically |
| 19:51.19 | seanbright | but without that it's slow as shit |
| 19:53.49 | *** join/#asterisk-dev CELYA_ (~Thunderbi@2a01cb0583d31d000459046f3d5f90e1.ipv6.abo.wanadoo.fr) |
| 19:55.03 | seanbright | gtjoseph: at one point i had a branch that i was testing and i believe i was using TLS and it was still pretty slow |
| 19:55.10 | seanbright | let me see if i can find it |
| 19:56.30 | gtjoseph | without all the locking? |
| 19:56.34 | seanbright | yes |
| 19:57.09 | seanbright | and i must have deleted it |
| 19:57.17 | seanbright | i think i discussed with file at some point |
| 19:57.23 | seanbright | let me check IRC logs and see if i can find it |
| 19:57.25 | file | I vaguely recall it |
| 19:57.34 | file | and I was probably sad and disappointed |
| 20:00.04 | seanbright | http://infobot.rikers.org/%23asterisk-dev/20171210.html.gz |
| 20:00.46 | file | ah that was the query |
| 20:02.16 | seanbright | http://infobot.rikers.org/%23asterisk-dev/20171211.html.gz |
| 20:02.21 | seanbright | continues the conversation a bit |
| 20:02.57 | file | oh, the dark time of me getting fed up with OPTIONS |
| 20:03.01 | seanbright | heh |
| 20:05.25 | seanbright | so what i gather from that mini-flashback is that the pthread locks weren't the problem, it was the commit "speed" |
| 20:06.09 | file | seanbright: what I was referring to was a sqlite3_open per thread using TLS above |
| 20:26.13 | seanbright | if you do a sqlite3_open you have to create a new prepared statement for that instance |
| 20:26.39 | file | yeah |
| 20:27.16 | seanbright | oh, but you'd only take the penalty once per thread |
| 20:27.18 | seanbright | i see |
| 20:28.19 | seanbright | i guess my contention is that we shouldn't have to work this hard |
| 20:28.29 | seanbright | we could have just upgraded bdb |
| 20:28.48 | seanbright | but spilled milk and all that |
| 20:35.50 | rmudgett | I thought there was a licensing issue with newer versions of bdb. |
| 20:39.37 | seanbright | > 5.3 is AGPL |
| 20:39.49 | seanbright | ubuntu ships 5.3 |
| 20:40.54 | seanbright | 5.3 is the sleepycat license |
| 20:41.20 | seanbright | we don't have to ship it though |
| 20:44.29 | seanbright | i was thinking leveldb, but it's "good" interface is c++ |
| 20:47.49 | *** join/#asterisk-dev zerohalo (sid288136@gateway/web/irccloud.com/x-dehktjsnbyrcawlb) |
| 20:47.50 | *** join/#asterisk-dev csavinovich (sid296765@gateway/web/irccloud.com/x-qjktlapyuqblkicl) |
| 20:47.51 | *** mode/#asterisk-dev [+o csavinovich] by ChanServ |
| 20:48.26 | *** join/#asterisk-dev moy (sid47040@gateway/web/irccloud.com/x-ovvvprhwgquitomu) |
| 20:51.03 | *** join/#asterisk-dev file (sid178970@asterisk/developer-and-muffin-lover/file) |
| 20:51.03 | *** mode/#asterisk-dev [+o file] by ChanServ |
| 23:36.03 | *** part/#asterisk-dev kharwell (kharwell@nat/digium/x-bubqplggmjjdhqoq) |