| 02:51.46 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:90b8:79c4:fa12:b04d) |
| 04:19.52 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:90b8:79c4:fa12:b04d) |
| 05:13.55 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:90b8:79c4:fa12:b04d) |
| 06:08.15 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:90b8:79c4:fa12:b04d) |
| 07:59.06 | *** join/#asterisk-dev AsteriskRoss_ (~AsteriskR@mx01.sfh.nor.gb.voicehost.co.uk) |
| 12:03.27 | *** join/#asterisk-dev sduthil (~sduthil@107-179-252-17.cpe.teksavvy.com) |
| 12:21.51 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:e16f:8b60:f318:f9fd) |
| 12:26.32 | *** join/#asterisk-dev AsteriskRoss_ (~AsteriskR@r01.nt-r1.nor.gb.voicehost.co.uk) |
| 13:24.29 | *** join/#asterisk-dev bford (uid283514@gateway/web/irccloud.com/x-wxekyelcmgytbbji) |
| 13:24.29 | *** mode/#asterisk-dev [+o bford] by ChanServ |
| 14:01.11 | *** join/#asterisk-dev kharwell (uid358942@gateway/web/irccloud.com/x-punkyfvygwzjnjae) |
| 14:01.11 | *** mode/#asterisk-dev [+o kharwell] by ChanServ |
| 14:24.32 | *** join/#asterisk-dev Juggie (~Juggie@unaffiliated/juggie) |
| 15:25.00 | seanbright | coronavirus has definitely helped with the number of crash reports we're getting |
| 15:25.10 | seanbright | and by helped i mean 'increased' |
| 15:25.55 | gtjoseph | :) |
| 15:33.15 | jkroon | seanbright, you think uptake in users? |
| 15:34.15 | seanbright | uptake in complainers |
| 15:34.18 | seanbright | just kidding. |
| 15:34.35 | seanbright | yes, i'm sure asterisk is getting exercised more than usual |
| 15:35.29 | seanbright | astridevcon 2020 (if it happens) topic: utf-8 handling |
| 15:35.44 | seanbright | might want to set aside a full day for that |
| 15:41.39 | file | oy |
| 15:42.35 | seanbright | does lvl come here? |
| 15:42.48 | file | I don't think so |
| 15:42.52 | gtjoseph | not sure |
| 15:53.56 | jsmith | seanbright: I'd be interested in your thoughts on internationalization improvements, even before AstriDevCon |
| 15:55.06 | seanbright | well, i think the ideal scenario would be that all C strings in asterisk were valid UTF-8 |
| 15:55.47 | seanbright | you enforce it at the edges |
| 15:56.26 | seanbright | and then we either make icu a required dependency, or we figure out what we can live with in terms of unicode support |
| 15:57.04 | seanbright | but every ast_copy_string() (where the destination is a fixed number of bytes) would need to be replaced with something multibyte aware |
| 15:57.21 | seanbright | it would be a disaster of epic proportions |
| 15:57.23 | seanbright | :D |
| 15:58.23 | jkroon | and all libraries used for string handling would need also need to be aware. |
| 15:58.27 | jkroon | eg, pjproject. |
| 15:58.48 | seanbright | eh... maybe |
| 15:59.11 | jkroon | is SIP itself capable of multibyte? |
| 15:59.16 | seanbright | SIP is utf-8 |
| 15:59.29 | seanbright | well... SIP is _specified_ to be UTF-8 |
| 15:59.37 | seanbright | but you know how that goes |
| 15:59.40 | jkroon | that solves that. |
| 15:59.52 | jkroon | anybody that crashes ... it's their bug :). |
| 16:00.08 | seanbright | i think that applies to the headers at least (and i'm sure SDP) |
| 16:00.14 | jkroon | but yea, experience dictates that it's always asterisk until conclusively proved otherwise. |
| 16:00.19 | seanbright | but if you carrying other content types that is up to them |
| 16:00.31 | gtjoseph | well, for most things it's just a passthrough. we can support utf-8 callerid for instance. |
| 16:00.56 | seanbright | unless the caller id is >= 80 characters |
| 16:01.10 | gtjoseph | well, other restrictions apply. :) |
| 16:01.12 | seanbright | and we truncate in the middle of an otherwise valid UTF-8 sequence |
| 16:01.22 | jkroon | again, if "expensive switch X" crashes due to UTF-8, and you're using asterisk, guess what's going to get blamed. |
| 16:01.58 | seanbright | that is what i said you enforce it at the endges |
| 16:02.00 | seanbright | edges* |
| 16:02.05 | gtjoseph | I think we can limit the scope to dislplan pattern matching and functions. |
| 16:02.22 | seanbright | so if something is coming in to asterisk, you make it utf-8 if it's not |
| 16:02.41 | jkroon | gtjoseph, then you need to be able to handle utf-8 patterns ... exten => _utf8-here,... |
| 16:02.51 | gtjoseph | exactly |
| 16:02.54 | jkroon | which means config file parsing also becomes a factor. |
| 16:02.58 | seanbright | if it is going out of asterisk, you can use, say, endpoint configuration, to strip urf-8 sequences or something if your endpoint can't handle them |
| 16:03.09 | gtjoseph | we can parse the config file correctly |
| 16:03.35 | jkroon | this sounds interesting and way outside of my experience. |
| 16:03.42 | gtjoseph | the issue is that we match byte by byte |
| 16:04.02 | seanbright | the issue is that we truncate |
| 16:04.19 | seanbright | if everything coming in to asterisk (conf, signaling, whatever) was valid UTF-8 |
| 16:04.28 | seanbright | and we never truncated anything |
| 16:04.29 | jkroon | gtjoseph, why is byte matching an issue? |
| 16:04.32 | seanbright | we'd be fine |
| 16:04.39 | gtjoseph | one sec. gonna paste an example |
| 16:05.12 | seanbright | if memory serves, we have a fixed length buffer for lines in config files |
| 16:05.26 | seanbright | so i can guarantee i can get broken UTF-8 into asterisk with config parsing as it is today |
| 16:05.34 | gtjoseph | Set(CALLERID(name)=${FILTER(Ð-Яа-ÑA-Za-z\040\050\051\055\056,${CALLERID(name)})}); |
| 16:06.08 | gtjoseph | If you notice the "Ð-Яа-Ñ" are Cryillic characters |
| 16:06.16 | seanbright | back up |
| 16:06.19 | jkroon | ah ok, in filters, yes, that makes sense. |
| 16:06.23 | seanbright | can you clarify what point you are making, gtjoseph? |
| 16:06.28 | seanbright | i may be confused |
| 16:06.43 | gtjoseph | my point is that even if we don't truncate we still don't match correctly |
| 16:07.01 | seanbright | ah, i see. that's fair. |
| 16:07.06 | jkroon | \x12\x12\x13\x14 ... right, so let's say that's two bytes per character, and you're looking to filter out \x12\x13 characters ... |
| 16:07.33 | jkroon | that should not match at all, but will, so end up with \x12\x14 instead of the original string. |
| 16:07.43 | seanbright | it's covered in the issue |
| 16:07.51 | seanbright | https://issues.asterisk.org/jira/browse/ASTERISK-28830 |
| 16:07.55 | gtjoseph | https://issues.asterisk.org/jira/browse/ASTERISK-28830 |
| 16:08.02 | file | dejavu |
| 16:08.08 | jkroon | this is for *18? |
| 16:08.09 | seanbright | all over again |
| 16:08.22 | seanbright | is what for 18? |
| 16:08.25 | gtjoseph | no it just came up as a bug today |
| 16:08.28 | jkroon | utf-8 |
| 16:08.33 | seanbright | god no |
| 16:08.37 | seanbright | asterisk 22 maybe |
| 16:08.38 | jkroon | oh crappy. |
| 16:09.07 | jkroon | ok ... right, so there is time. cause i can just imagine how much chaos implementing that is going to create. |
| 16:09.12 | seanbright | if we had to make icu a required dependency that would take at least 2 released i'd imagine |
| 16:09.42 | jkroon | longs for the days where latin-1 was the most complex situation you had to face. |
| 16:09.43 | seanbright | and then someone would want to make the unicode handling pluggable |
| 16:09.51 | jkroon | :) |
| 16:09.53 | gtjoseph | i think this is getting more complicated than it needs to be. |
| 16:09.53 | seanbright | so we'd have to create a layer of abstraction |
| 16:10.04 | seanbright | gtjoseph: nice, so how do we fix? |
| 16:10.44 | gtjoseph | for this particular issue, we could have FILTER take an encoding argument |
| 16:10.49 | seanbright | ah, yes, i see |
| 16:10.54 | seanbright | for this particular issue |
| 16:11.28 | gtjoseph | we still have truncation issues yes |
| 16:11.48 | gtjoseph | but for the most part, we can just passthrough |
| 16:11.59 | seanbright | indeed, i was being silly |
| 16:12.01 | seanbright | carry on |
| 16:12.36 | seanbright | jsmith: we're good. going to add a charset argument to FILTER |
| 16:12.46 | seanbright | problem. solved. |
| 16:13.04 | gtjoseph | seanbright: are you taking this issue? |
| 16:13.07 | seanbright | nope |
| 16:13.17 | seanbright | because that is an awful solution |
| 16:13.21 | seanbright | (no offense) |
| 16:13.25 | gtjoseph | oh, you're just being _you_ then. |
| 16:13.42 | seanbright | let's modify a single string handling dialplan function to take an encoding argument |
| 16:13.55 | seanbright | that is the solution? |
| 16:14.32 | gtjoseph | in places where we have to do pattern matching |
| 16:14.43 | seanbright | k |
| 16:14.54 | gtjoseph | we could set it in asterisk.conf, or make it a dialplan global variable |
| 16:15.07 | gtjoseph | we already have language in asterisk.conf |
| 16:15.24 | seanbright | great, sounds awesome |
| 16:20.19 | gtjoseph | seanbright: if I pay you for "remote-hands" work, would you please slap yourself? |
| 16:20.53 | seanbright | one of us is being paid to be here |
| 16:20.55 | seanbright | and it's not me |
| 17:32.51 | jkroon | bumped into yet another issue from the musl team @ Gentoo. https://wiki.musl-libc.org/faq.html#Q:-When-compiling-something-against-musl,-I-get-error-messages-about-%3Ccode%3Esys/cdefs.h%3C/code%3E |
| 17:33.29 | jkroon | I'm happy to patch, and provide a submission. just need to check if everyone is happy with the approach described there? |
| 17:35.45 | seanbright | what file? |
| 17:36.15 | jkroon | astdb2sqlite3.c:39 |
| 17:36.23 | jkroon | i did a quick grep, there are a few others too. |
| 17:37.43 | jkroon | i've got a musl chroot here which I used to fix up dahdi. so I can probably track them all down there. |
| 17:38.01 | seanbright | god speed |
| 18:26.29 | coreyfarrell | I actually half agree with Alexander about not committing the output from bootstrap, this is something about asterisk repo I've always found strange. that said I don't see it as such a big deal as he seems to feel. |
| 18:27.29 | file | committing it at all? |
| 18:31.14 | coreyfarrell | what would make sense to me is for everything generated by bootstrap would be listed in .gitignore. have the release process bootstrap and include bootstrap output in the official source tarballs. |
| 18:32.09 | file | nods |
| 18:33.42 | jkroon | i agree on that as well. |
| 18:33.56 | jkroon | for what that is worth. |
| 18:34.54 | jkroon | seanbright, you mentioned asterisk is being more used ... yet we saw around a 65% drop in volumes w.r.t. minutes/day. |
| 18:35.15 | seanbright | then i must be wrong |
| 18:35.23 | seanbright | thanks for letting me know |
| 18:36.16 | jkroon | could be that people are trying different things now to accommodate lockdowns and different scenarios and happening onto issues. i know we are trying a lot of different stuff at the moment. |
| 18:36.42 | jkroon | and if they found it previously it was just like "oh well", whereas now there's pressure on them to get it fixed. |
| 19:14.38 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:11f6:aca2:9efe:7efc) |
| 19:25.19 | *** join/#asterisk-dev pchero_work (~pchero@2a02:a210:2241:6480:11f6:aca2:9efe:7efc) |
| 23:45.13 | *** join/#asterisk-dev jkroon (~jkroon@165.16.203.106) |