**** BEGIN LOGGING AT Thu Jul 19 03:00:02 2018 Jul 19 05:11:33 gored moaning Jul 19 05:16:58 Morning people! Jul 19 06:22:29 mmm Jul 19 06:22:44 wtf arch linux Jul 19 06:23:01 trying to set up a vm to reproduce the imagebuilder issues and cannot even get an ssh server running Jul 19 06:23:09 so if I have libs in /usr/lib/X, how can I tell them to look there at runtime? Jul 19 06:23:20 already spent way more time on this than I actually wanted Jul 19 06:23:26 jow: pacman -Syu openssh && systemctl enable sshd && systemctl start sshd Jul 19 06:23:29 :D Jul 19 06:23:34 also wtf pacman Jul 19 06:23:55 pacman install foo -> no operation check help Jul 19 06:24:09 pacman --help -> no single mention of the word "install" Jul 19 06:24:20 its called --sync... right Jul 19 06:24:24 and people complain about openwrt Jul 19 06:24:28 -S (install), -y (sync repos), -u (update system) Jul 19 06:25:11 official instal lguide mentions one should install a boot loader, but neither mentions the actual pacman isntall command nor the package name to use Jul 19 06:25:31 is it really required to use this weird bootstrap live iso thing or are there ready live isos or something? Jul 19 06:25:49 the iso is a live cd, you could use that Jul 19 06:25:54 if you have enough ram, or a disk to mkfs Jul 19 06:27:08 whats not working with the imagebuilder? Jul 19 06:27:19 if its not kernel related, you could just use a chroot Jul 19 06:27:27 no idea, some arch linux snowflake thing I guess Jul 19 06:27:34 is there an issue? Jul 19 06:27:38 or is it on list? Jul 19 06:27:41 yeah, glibc assertions Jul 19 06:27:51 probably needs some more beating into submission Jul 19 06:28:01 about to find out Jul 19 06:28:09 gimme a url, I'll check Jul 19 06:28:16 my build container is arch Jul 19 06:29:38 https://forum.lede-project.org/t/solved-by-using-ubuntu-instead-or-arch-lede-imagebuilder-17-01-5-ar71xx-generic-linux-x86-64-inconsistency-detected-by-ld-so/17117/5 Jul 19 06:32:43 I see that error too Jul 19 06:32:59 [jwh@aurbuilder lede-imagebuilder-17.01.5-ar71xx-generic.Linux-x86_64]$ pacman -Q glibc Jul 19 06:33:02 glibc 2.27-3 Jul 19 06:35:02 seems others have stumbled upon it too Jul 19 06:35:39 maybe systemd related Jul 19 06:39:49 no, I know what it is Jul 19 06:40:02 o Jul 19 06:43:37 bingo Jul 19 06:43:45 forgot to backport a patch Jul 19 06:43:51 doh Jul 19 06:43:55 sed -i.bak -re 's#/(usr|lib|bin|etc)#/xxx#g' ./staging_dir/host/lib/libc.so.6 Jul 19 06:43:58 this will fix it Jul 19 06:44:23 looks good Jul 19 06:44:29 jow: Are MS_NOATIME parameters dependcies for "something" else than OS itself? https://git.openwrt.org/?p=project/procd.git;a=blob;f=initd/early.c;h=00fd9467912085fcba2b2abab2163e49b025009e;hb=HEAD . There is a fix for lxc https://github.com/lxc/lxc/pull/2480/commits/69eadddb37352ff29fd785577 which will enable unprivileged LXC containers with NOA Jul 19 06:44:30 TIME parameter. Question is.. Patch procd, or patch lxc ? Jul 19 06:44:39 muhaha: procd Jul 19 06:45:37 -rw-r--r-- 1 jwh jwh 16252928 Jul 19 06:44 lede-17.01.5-ar71xx-generic-archer-c7-v2-squashfs-factory.bin Jul 19 06:45:44 guess it works Jul 19 06:45:45 heh Jul 19 06:47:35 jow: Is this patch ok? https://pastebin.com/DVPu0gXZ Jul 19 06:51:28 given that these all are virtual fs'es I would say yes, but thats just an uninformed opionion, not sure if it has any potential consequences Jul 19 06:51:47 who can tell? :) Jul 19 06:52:35 so... what magic do I need to load libs from a specific dir : Jul 19 06:53:03 Hello. What would cause everytime i go to the web interface, Luci tries to get me to download the corresponding .bin file? Jul 19 06:53:09 I mean, is rpath the only way, or can I just add a path to the search dirs Jul 19 06:53:12 ? Jul 19 06:53:31 LD_LIBRARY_PATH Jul 19 06:53:58 mm, can't really expect people/scripts/whatever to do that Jul 19 06:54:02 Bukowski: uhm what? Can you elaborate some more? Do you mean the browser offers you a binary file download? Jul 19 06:54:13 jwh: what do you intend to do? Jul 19 06:54:16 well Jul 19 06:54:25 ship some nonstandard libs with a program? Jul 19 06:54:28 jow: Exactly. If I go to 192.168.1.1 ut tries to get me to download luci.bin Jul 19 06:54:40 the shared libs are in /usr/lib/accel-ppp, I *could* put them in /usr/lib I guess Jul 19 06:55:01 Bukowski: ah - first of all; which version of openwrt or lede is installed? Jul 19 06:55:13 jwh: yes, that would be rpath Jul 19 06:55:16 but... there is a remote possibility that there might be other libs named the same Jul 19 06:55:25 yeah Jul 19 06:55:36 probably the easiest way Jul 19 06:56:23 version 17.01.4 Jul 19 06:56:46 and its LEDE Jul 19 06:57:37 this seemed to have happened right after installing packages for sensors. i followed this tutorial: https://wiki.openwrt.org/doc/howto/luci_app_statistics and the next thing i knew i was getting .bin file download requests Jul 19 06:57:40 ssh works fine though Jul 19 06:58:31 Bukowski: can you via ssh execute /www/cgi-bin/luci and check if it outputs html Jul 19 06:58:36 or if it prints an error message? Jul 19 07:00:08 it prints html Jul 19 07:01:06 Bukowski: okay, now please run "wget -O - http://127.0.0.1/cgi-bin/luci" and tell me if it prints html too Jul 19 07:01:12 (again, via ssh) Jul 19 07:04:07 yes it prints html as well Jul 19 07:04:37 Bukowski: what about "wget -O - http://127.0.0.1/luci" ? Jul 19 07:05:34 Bukowski: also can you pastebin the output of "wget -O - http://127.0.0.1/cgi-bin/luci | head -n 10" ? Jul 19 07:05:38 404 error Jul 19 07:05:58 wait, actually Jul 19 07:06:06 Where are these files downloading to? Jul 19 07:06:14 stdout (-O -) Jul 19 07:06:31 so they're just printed and not stored Jul 19 07:07:06 bleh, rpath upsets dependency checks Jul 19 07:07:07 :D Jul 19 07:07:48 okay it seems like it downloaded the first one, because there was a loading message. /cgi-bin/luci Jul 19 07:09:43 but yea just /luci gives 404 error Jul 19 07:09:59 thats okay, just want to rule out problems Jul 19 07:10:20 so yeah, can you copy the output of "wget -O - http://127.0.0.1/cgi-bin/luci | head -n 10" into pastebin.com and send it to me? Jul 19 07:12:56 https://pastebin.com/r4nSrhWR Jul 19 07:14:02 Bukowski: funny, so it is not actually executing luci but offering its source code for download Jul 19 07:14:33 Bukowski: so next things to check are (please pastebin again): ls -lh /www; ls -lh /www/cgi-bin; ls -lh /www/cgi-bin/luci Jul 19 07:14:44 they all should be executable and have world read permissions Jul 19 07:17:43 Bukowski: and finally I need to know the output of "ps ww | grep uhttpd" as well Jul 19 07:19:41 https://pastebin.com/CbmNFz2G https://pastebin.com/RCmP58KS Jul 19 07:19:46 thank you for the help by the way Jul 19 07:21:13 hmm Jul 19 07:22:50 would it have anything to do with collectd enable or lucy_statistics enable ? thats the last thing i did Jul 19 07:23:45 it sohuldn't Jul 19 07:24:16 nothing is impossible, my hunch was that one of the packages might be broken, somehow messing with system permissions Jul 19 07:26:59 Bukowski: that first pastebin you sent, was that really the output of (exactly!) "wget -O - http://127.0.0.1/cgi-bin/luci | head -n 10" ? Jul 19 07:27:08 I would've expected at least ten lines of output Jul 19 07:29:20 yes, that was the entire output, i wrote the output to a text document just in case and its the same Jul 19 07:29:47 oh you know, i missed one line Jul 19 07:30:16 luci.dispatcher.indexcache = "/tmp/luci-indexcache" Jul 19 07:30:16 and if you run "cat /www/cgi-bin/luci" on the command line, its also just the three or four lines? Jul 19 07:30:25 or way longer? Jul 19 07:30:51 its the same Jul 19 07:31:14 thats not sounding good Jul 19 07:31:58 ah nvm, the last line should be "luci.sgi.cgi.run()" Jul 19 07:32:06 its really just 6 lines long or so Jul 19 07:32:46 oh yeah, that was hanging behind the root@LEDE Jul 19 07:33:01 i also installed lightppd today, could that cause a problem? Jul 19 07:37:18 Bukowski: hehe... definitely :D Jul 19 07:37:31 oh great, lol Jul 19 07:37:37 Bukowski: can you please run this command: "(echo -en "GET /cgi-bin/luci HTTP/1.0\r\n\r\n"; sleep 1) | nc 127.0.0.1 80 | head -n 10" and pastebin the output again? Jul 19 07:37:48 That should dump the http headers (and tell us which web server is responding) Jul 19 07:41:11 looks like lighttpd https://pastebin.com/Rd1hMPAG Jul 19 07:41:40 so that explains your issue - lighthttpd is not configured to serve luci as cgi-program Jul 19 07:41:52 it simply offers the script source code for downloading Jul 19 07:42:18 ohh. can i set it up to do so? Jul 19 07:42:24 jow: so procd managed processes need to stay in the foreground right? Jul 19 07:42:31 Bukowski: https://wiki.openwrt.org/doc/howto/luci.on.lighttpd Jul 19 07:42:36 jwh: yep Jul 19 07:42:39 nice, thank you so much jow Jul 19 07:42:39 kk Jul 19 07:42:44 jow: haha, funny. everybody in my team uses arch Jul 19 07:42:48 I don't Jul 19 07:42:51 putting the finishing touches on this package Jul 19 07:43:01 I think its working Jul 19 07:43:03 maybe Jul 19 07:43:22 syslog doesn't actually seem to write to syslog Jul 19 07:43:25 so that doesn't help Jul 19 08:36:20 Is possible that I am getting patch hunk failed error because of patch was created from master and I am applying it to specific checkout ? Nothing is changed of course between master and specific sha, but I have some metadata generated from git patch Jul 19 08:41:15 For example: mkdir -p ./package/system/procd/patches/ && cp ./package/system/procd/patches/my.patch && make tools/install && make toolchain/install && make -j1 V=s package/procd/compile Jul 19 08:41:35 Is it correct? Jul 19 08:42:03 This is the patch https://pastebin.com/DVPu0gXZ Jul 19 08:56:10 muhaha: afaik the git diff metadata is only used if it cannot apply the patch directly...it then does this 3-way merge thing using the commit refs contained in the patch. Jul 19 09:06:56 muhaha: You would also have more success with answering the 'any potential consequences' question by submitting to the email list instead of just posting here as a pastebin on IRC. Jul 19 09:10:51 blogic: might I suggest using the github 'description' field to clearly state that it's a second-class citizen when it comes to sending patches and that the m-l should be preferred? This is stated nowhere... Jul 19 09:17:51 Hi Jul 19 09:18:30 In our project ipk package name changed from 1.ipk to 2.ipk . How do i handle this name change ? Jul 19 09:18:40 For rpms we have Package and Obsolete tags. Which we can use Jul 19 09:18:47 But for ipks how do i do it ? not able to find anything from internet Jul 19 09:19:34 define "handle"? Jul 19 09:27:04 f00b4r0: no Jul 19 09:27:29 f00b4r0: i'd rather have the less skilled folks hang out there and pull the more skilled ones silently to the ML Jul 19 09:27:47 heh Jul 19 09:28:57 so when i install 2.ipk my older version of it i.e. 1.ipk should get un-installed Jul 19 09:30:04 Basically i want to obsolete my .ipk when i install my .ipk Jul 19 09:31:24 Looking for an equivalent of "Obsolete" tag for rpms Jul 19 09:34:05 * ldir tries to throw oil on troubled waters. Jul 19 09:56:30 ldir: here have a lighter Jul 19 09:57:10 * ldir screams - where have my eyebrows gone? :-) Jul 19 10:18:21 jow: is the 18.06 release signed with a different key than 17.01? Jul 19 10:20:36 nevermind, found it in the wiki https://openwrt.org/docs/guide-user/security/signatures?s[]=gpg Jul 19 10:51:38 blogic: were you able to reproduce my issue? Jul 19 11:03:37 I'll leave this here whilst having lunch - Now that cake has gone upstream I'm thinking it would be a good idea to drop the separate cake kmod package and backport it as a kernel patch. Discuss? Throw bricks etc :-) Jul 19 11:30:33 ldir: +1 Jul 19 11:41:10 ldir: i think it depends on how often you expect to update it Jul 19 11:41:47 if you expect to continue updating it frequently, backport patches are probably more work Jul 19 11:41:51 and maybe more effort to test Jul 19 11:42:18 if there aren't going to be too many changes, this seems like a good idea Jul 19 11:58:09 hmmm I forgot again how to debug a process that doesn't start in procd Jul 19 11:58:26 it runs fine when I do it manually, no logging whatsoever when run through procd Jul 19 11:58:35 PROCD_DEBUG=1 is not wat I was looking for Jul 19 12:00:00 nbd: well naively I don't expect any (minimal) changes, since it has at long last gone upstream. Jul 19 12:01:10 INIT_TRACE=1 Jul 19 12:01:56 hah, forgot command in procd_append_param Jul 19 12:08:39 nbd: it can't be an exact backport of the upstream commits of course because of different kernel versions. And it's a certain amount of work, but I'd really like to get rid of the separate package. Jul 19 12:31:21 holy shit. git send-email won't save email file if it fails to send Jul 19 12:31:40 i give up. Jul 19 12:33:05 go do something else for a day or two... then retry. I tend to save my commit message window just in case. Jul 19 12:33:29 no, i'm done. Jul 19 12:33:41 ok. Jul 19 12:33:43 this is 2018, not 1975. Jul 19 12:40:48 Hello Jul 19 12:41:16 I need some help with OpenWRT, is anyone online? Jul 19 12:51:01 hey Jul 19 12:51:52 just ask your question Jul 19 12:54:37 Hi guys! Short question. My internet connection seems to be screewed up lately. I am running a LEDE router behind my ISPs modem. Occasionally, "some part" of the internet becomes unavailably to reach. I can still browse Google or use my TeamViewer connection. But other site are becoming unavailable. Using the diagnostics section within LEDE, I can ping google, but at the moment I cannot ping tomshardware.com for example. 100% pack Jul 19 12:54:49 When I reboot the router, everything will work perfectly for a while. Jul 19 12:55:13 This happes 3-4 times a day, I guess... any ideas on how to troubleshoot further? Thanks!! Jul 19 12:55:23 jwn: traceroute Jul 19 12:55:30 to see where the problem begins Jul 19 12:55:53 as you mention 100% packet loss, that means it can resolve the IP so your connection is OK, as DNS can be resolved Jul 19 12:57:32 jwn: i can't ping it either - they're filtering icmp Jul 19 12:59:30 stintel: Thanks! Tracerouting dev.openwrt.org: 10 hops. Tracerouting tomshardware.com or linux.com: infinite hops, as it seems. 10 hops, then only "*", "*", etc... Jul 19 13:00:30 jwn: pastebin? Jul 19 13:00:36 m4t: You're correct. Still, I cannot open websites on that domain. Also, traceroute looks quirky. Jul 19 13:01:43 Cannot open pastebin.com :-D. Wait, Im connected from my work PC via Teamviewer, will paste shortly. Jul 19 13:01:48 :D Jul 19 13:02:26 stintel: https://pastebin.com/5922f7kL Jul 19 13:04:01 can you try traceroute -I ? Jul 19 13:04:41 and also dns resolution would be cool Jul 19 13:05:40 Okay, this is from the route via ssh: https://pastebin.com/htM6iABt Jul 19 13:05:46 but looks like it's a problem at ATT Jul 19 13:05:58 NTT* Jul 19 13:06:02 Hello everyone. I need to revert back my 840N v4 to original TP-LINK firmware. So I downloaded the original firmware from TP-Link site, rename as tp_recovery.bin and did the upload it using tftp. After this my router is rebooting. I tried to put back the OpenWrt and the problem is the same. What I need to do to have my router back? :) Jul 19 13:06:09 cuz I can reach linux.com fine via traceroute Jul 19 13:06:32 jwn: send traceroute outputs to your ISP and complain Jul 19 13:06:53 stintel: You think it's my ISPs issue? Jul 19 13:07:07 I bet you, if I restart my router now, it will work :-) Jul 19 13:07:18 jwn: that would be weird Jul 19 13:07:21 but try it Jul 19 13:07:25 just a sec Jul 19 13:10:06 * ldir is about to expose his complete lack of patch skill.......... but the only silly question is the one you didn't ask..... so...... Jul 19 13:11:37 stintel: Restart done. I can access tomshardware and linux.com without problems. Traceroute still looks the same, though... Jul 19 13:11:54 huh Jul 19 13:12:21 Yup. Sites are now working without problems. Jul 19 13:12:44 Could this be an IPv4 / IPv6 configuration issue on my site? Jul 19 13:12:49 I have a patch to a source tree that has hunks, some of which don't apply. I get a rejects file. Sometimes it's a large section of text that's been rejected, many lines, but it doesn't apply because of a lack of correct context. If I had the file I'd just cut'n'paste or whatever the new stuff in... buuuut the rejects file has a + at every line. There must be a sensible way to deal with this? Jul 19 13:13:08 jwn: did you use traceroute -I (to force use of ICMP instead of UDP)? Jul 19 13:13:18 yes, I did Jul 19 13:13:20 with UDP I am getting the same as you Jul 19 13:13:38 what traceroute program is this? Jul 19 13:14:37 ah, if it's the busybox one it doesn't have the -I option so it's still using udp Jul 19 13:15:05 I am connected to my router via ssh and using traceroute from console. Jul 19 13:15:13 yeah ok that doesn't support ICMP :( Jul 19 13:15:44 Hello, I need layer 2 bridging instead layer 3, how can I achieve that? Jul 19 13:15:47 oh. alright. Still, any ideas on why I cannot browse certain sites until after a reboot? Jul 19 13:16:09 jwn: next time it happens, check dmesg and logread for pointers Jul 19 13:16:14 I aint even rebooting my ISP's box, just the router I put behind it Jul 19 13:16:17 maybe a conntrack problem Jul 19 13:16:39 dmesg via ssh? Jul 19 13:16:57 Or somethig on Luci? Jul 19 13:17:01 yes Jul 19 13:17:07 Okay, will do! Jul 19 13:20:08 Thanks, stintel! Jul 19 13:21:49 wow. I just received an onslaught of DMARC failure report messages after posting to the m-l Jul 19 13:22:24 all these silly antispam measures Jul 19 13:22:29 coming from 'abuse@163.com' Jul 19 13:26:00 @f00b4r0 always been an issue with China but dont really blame them, no youtube and a proper strict country. Jul 19 13:26:20 more then likely scanning for open proxy,vpn's as such Jul 19 13:52:40 * ldir is scared off the backport cake idea - cake is easy... it's the nf_conntrack hooks that are giving me the willies. Jul 19 14:25:54 jwn: ppp involved anywhere? Jul 19 14:28:17 jwn: https://www.revk.uk/2013/11/bt-huawei-fttc-modem-bug-breaking-vpns.html & https://www.revk.uk/2018/07/very-frustrating-modem-bug.html Jul 19 14:30:12 ldir: a chinese modem breaking vpns? one could think this is intentional ;> Jul 19 14:37:19 heh, I have a huawei fttc modem Jul 19 14:37:26 mmm, is there anything that can monitor say the defaulr route and trigger an action if it changes? Jul 19 14:37:57 fortunately I *can* use VPN :) Jul 19 14:38:35 god damn this pile of shit Jul 19 14:38:53 I don't think I have shared library problems now, but its still complaining Jul 19 14:39:15 oh confusing Jul 19 14:39:20 we have jwn and jwh Jul 19 14:39:39 namespace collision, I was here first Jul 19 14:39:39 :D Jul 19 14:40:37 whaaaat Jul 19 14:40:44 moaning about not being able to find symbols in libraries Jul 19 14:40:48 *why* Jul 19 14:40:51 it knows they're there Jul 19 14:45:20 gninrom Jul 19 14:54:28 * ldir whimpers - well this has gone well then - https://pastebin.com/1XJ8kj02 Jul 19 14:54:45 :( Jul 19 14:55:48 jwh: have you tried --stop-complaining-look-harder option? Jul 19 14:56:34 lol Jul 19 14:56:46 well its stopped moaning about being unable to load libs or find symbols Jul 19 14:57:00 but now its logging "not found" for a bunch of config options Jul 19 14:57:07 which I suspect is the same thing Jul 19 14:57:36 mrs ldir would invoke with --have-you-tried-lifting-a-few-things-up Jul 19 14:57:43 lol Jul 19 14:57:54 why do people write horrible code Jul 19 14:57:56 or randomly retry the same thing a few times Jul 19 14:58:10 I mean I write horrible code but thats because I don't know any better Jul 19 14:59:30 maybe its coz they're russian and their english sucks Jul 19 14:59:41 sacre bleu Jul 19 15:01:17 uhhhhhhhhhhhh Jul 19 15:01:23 strcpy(fname, path); Jul 19 15:01:23 strcat(fname, "/lib"); Jul 19 15:01:23 strcat(fname, opt->name); Jul 19 15:01:23 strcat(fname, ".so"); Jul 19 15:01:24 christ Jul 19 15:01:26 what is that Jul 19 15:01:50 it's totally fine! nothing to see here! Jul 19 15:02:37 so not only does it expect the linker to know where they are, but it seperately loads them Jul 19 15:02:41 what the duece Jul 19 15:03:46 I think this is just my lack of cmake clue really Jul 19 15:03:55 I give it rpath but it seems to ignore it Jul 19 15:05:01 jwh: wasit you who went crazy with some SFP gear? Jul 19 15:05:10 yes Jul 19 15:05:36 jwh: I need some help to understand some basics, can we switch to query? Jul 19 15:05:40 sure Jul 19 15:12:47 18.06 tomorrow...? Jul 19 15:13:09 jow: ping (for buildbot, if you have time) Jul 19 16:09:28 Package suricata is missing dependencies for the following libraries: libevent_pthreads-2.0.so.5 Jul 19 16:09:29 .config:# CONFIG_PACKAGE_libevent2-pthreads is not set Jul 19 16:09:37 hehe Jul 19 16:10:36 stintel: i might be wrong but those dmarc error messages look like a configuration problem on the mailing-list side... Jul 19 16:11:11 it appears to be mangling original sender's dkim signature Jul 19 16:32:21 common problem with the way lists work Jul 19 16:33:10 indeed. Jul 19 16:33:17 can be mitigated though Jul 19 16:33:34 well thats why the list wraps the original message and adds that text Jul 19 16:33:45 arguably there is no need for the boilerplate everytime though Jul 19 16:34:03 if it worked correctly i wouldn't receive ruf reports I believe Jul 19 16:34:47 ah hm Jul 19 16:35:04 maybe its leaving the original header intact and people/software are parsing it Jul 19 16:35:11 looks that way Jul 19 16:36:01 f00b4r0: can I move the kmod-usb3 from Device/MikroTik to Device/mikrotik_rbm33g? Jul 19 16:36:23 f00b4r0: it looks like the package was left out intentional for the rbm11g Jul 19 16:36:51 mkresin: you'd have to ask tobleminer-tSYS Jul 19 16:37:03 sigh Jul 19 16:37:06 then again these devices have sizeable flash Jul 19 16:37:12 I am totally not happy with the amount of patches I have to make this crap work Jul 19 16:37:19 going to be an absolute nightmare Jul 19 16:37:27 f00b4r0: ehm. it's your patch which changes the default package selection for the rbm11g Jul 19 16:37:32 (musl is partly to blame, though) Jul 19 16:37:32 hold on Jul 19 16:37:59 i can answer the question Jul 19 16:38:23 the rbm11g has a mini-pcie slot, like its bigger sibling Jul 19 16:38:36 the usb mod is necessary for usb-over-pcie support (which is mandatory for a lot of LTE modules) Jul 19 16:38:50 that's why I grouped the kmod-usb for both Jul 19 16:39:04 f00b4r0: makes sense Jul 19 16:39:27 f00b4r0: fine if I extend the commit message to point out the change was intentional? Jul 19 16:39:44 sure Jul 19 16:40:02 mkay. will take care of the patch Jul 19 16:42:17 holy sh*t Jul 19 16:42:23 send-email is _really_ finicky Jul 19 16:42:31 i thought i got it to work and it broke again Jul 19 16:42:45 its all just horrible Jul 19 16:42:50 I manage to fuck it up every time Jul 19 16:42:52 even with noes Jul 19 16:42:52 and another 20mn of editing the cover message are gone Jul 19 16:42:55 notes* Jul 19 16:42:58 i am *fuming* Jul 19 16:43:31 git format-patch --cover-letter -o outgoing Jul 19 16:43:38 then edit the files with your favorite editor Jul 19 16:43:42 then git send-email outgoing/* Jul 19 16:43:52 stintel: my savior Jul 19 16:43:57 thanks! :) Jul 19 16:44:09 ;) Jul 19 16:44:36 * f00b4r0 tries again Jul 19 16:44:53 (and tries to remember the exact content of his cover) Jul 19 16:46:14 fml Jul 19 16:46:19 https://i.imgur.com/nY6YoQI.png Jul 19 16:46:25 gonna have to do something else before I smash something Jul 19 16:47:19 needs more strace Jul 19 16:52:54 DonkeyHotei: ping Jul 19 16:57:27 this software is definitely 100% garbage Jul 19 16:57:39 I mean who implements their own method of loading libraries Jul 19 16:59:37 hmm is it using dlopen()/dlsym()? that's standard Jul 19 16:59:46 maybe try an ltrace to see what it's doing Jul 19 16:59:54 its just garbage, check this... Jul 19 17:00:37 it uses dlopen, but it does a ridiculous dance beforehand Jul 19 17:01:19 https://github.com/xebd/accel-ppp/blob/master/accel-pppd/triton/loader.c#L40 Jul 19 17:04:03 pretty sure the path nonsense is tripping me up Jul 19 17:05:07 annoyingly it doesn't exit or anything when it fails to load stuff Jul 19 17:09:30 honestly I'm tempted to just patch it to make sure everything is linked Jul 19 17:09:34 rather than loading on demand Jul 19 17:10:03 junk that loader entirely Jul 19 17:19:20 jwh: what is this thing that's broken? Jul 19 17:19:32 lol, so its loading libs in the wrong order too since it has no dependency awareness Jul 19 17:19:36 g fucking g Jul 19 17:19:58 russell--: in-kernel pptp/l2tp/ipoe/sstp "server" Jul 19 17:20:15 the userland daemon is embarassing though Jul 19 17:21:18 Good day. I am new to openwrt and have tried to set up openwrt to act a wifi bridge, enabling non-wifi devices to connect to my wifi. However, I am having some trouble and was asking if I may ask for some support here. Jul 19 17:21:28 libpptp contains implicit references to functions in other modules, which aren't in the default config, and since it has no dependency stuff, it doesn't load them Jul 19 17:21:31 * jwh rollseyes Jul 19 17:23:29 rejigging the config does the trick though Jul 19 17:24:25 buxe2000: probably better to ask in #openwrt for usage problems Jul 19 17:24:46 nobody should be using pptp anyway Jul 19 17:24:50 so perhaps just not bother Jul 19 17:25:08 rusell-- thanks, will ask there! Jul 19 17:25:14 SSTP is the only way forward for crappy road warrior type stuff Jul 19 17:31:55 russell--: https://github.com/joeholden/staging-packages/tree/master/accel-ppp Jul 19 17:43:02 aaaand now its not showing up in menuconfig again Jul 19 17:43:02 wtf Jul 19 17:43:35 oh Jul 19 17:43:55 think I got my depends wrong Jul 19 17:47:52 need to select rather than depend on Jul 19 17:48:18 rmilecki: Another user in #openwrt just reported https://bugs.openwrt.org/index.php?do=details&task_id=1647 again with another device (Asus RT-N16). I did add some output from the serial console to the bug report. Jul 19 17:51:52 how is everyone finding the "Software Flow Offloading" ? Jul 19 17:52:20 and is it in ALL devices or certan ones for me? Jul 19 17:52:36 for now sorry** Jul 19 17:52:40 last time I enabled it my router started randomly rebooting Jul 19 17:53:10 i see Jul 19 18:01:44 mkresin: lack of reply-all unintended. Fixing. Jul 19 18:03:56 f00b4r0: hold on Jul 19 18:04:08 f00b4r0: you got my comment wrong Jul 19 18:04:10 did i break something? Jul 19 18:04:31 f00b4r0: no, I'm talking about apples and you are understand peaches Jul 19 18:04:39 ? Jul 19 18:04:48 f00b4r0: my point isn't that you are introduced fix-partitions. In fact, I welcome the change. Check the lantiq commit history, I've done it for lantiq on my own. Jul 19 18:04:55 f00b4r0: It is that you dropped the label and not using the "partitions" keyword to get the compiler accepting your overlapping partitions. My example above just outlines what I'm the opinion overlapping partitions should be expressed the devicetree way. Jul 19 18:05:05 is there something I can put in the procd init file to tell procd to kill a process with a specific signal? Jul 19 18:05:16 the label is unecessary Jul 19 18:05:25 it's taken from the node name when undefined Jul 19 18:05:32 f00b4r0: I know, not strictly required Jul 19 18:05:34 no need to duplicate text Jul 19 18:05:52 f00b4r0: but that is what I meant with common style, different than all other dts in that target Jul 19 18:06:00 ah Jul 19 18:06:09 well, I'm following documentation :P Jul 19 18:06:30 and it looks I'll have to re-reply to your other email as the mailing list is unhappy with my redirected message Jul 19 18:06:51 f00b4r0: I know, but I've the whole target in mind. hence I've for sure different requirements than you Jul 19 18:07:39 mkresin: I'm of the opinion that "because something has been done in a suboptimal way out of habit" is not exactly a valid reason to keep doing it that way... Jul 19 18:07:42 maybe we disagree on that Jul 19 18:08:12 unless there's some obvious reason to use a separate label which I'm not seeing Jul 19 18:08:14 f00b4r0: maybe we agree on that :-) Jul 19 18:08:15 (very possible too) Jul 19 18:08:46 f00b4r0: it is just that I would like to see some common sense at least for the whole target. even better for the whole tree Jul 19 18:09:02 well sure. But I see where this is going again Jul 19 18:09:07 you're trying to shoot the messenger Jul 19 18:09:21 f00b4r0: it's hard to explain if you didn't failed or introduced some breakage because something was done slightly different for one target Jul 19 18:09:23 i.e. the first person who tries to do it "right" is bound to fix everything else or get NAK'd. Jul 19 18:09:29 at least that's how it feels to me. Again. Jul 19 18:10:09 f00b4r0: no, it isn't that way. grep through the git history and check some of my tree/target wide changes Jul 19 18:10:42 f00b4r0: I feel your pain, as I've tried to fix stuff treewide and I know how much work it is Jul 19 18:11:49 mkresin: minor emergency have to bbiab. Jul 19 18:11:55 f00b4r0: and you are not the first who is asked to do so: https://git.openwrt.org/c9c4b2116c0984eb8baeff03e806030ee9c6175f Jul 19 18:12:47 Hi guys / girls Jul 19 18:13:05 I have a noob question to ask about Meraki MR18 APs if anyone can help Jul 19 18:14:08 f00b4r0: to summarize, neither I have a personal problem with you, nor do ask to do stuff which I havn't asked other contributer before Jul 19 18:30:16 mkresin: i don't have a personal problem with you either. However I am not interested in 'personal preferences': decisions should be technically argumented whenever feasible. As for the second part, just because some people don't mind doing your bidding doesn't mean everybody does. I'm not interested in being coaxed into fixing _everything_ when I offer to fix _something_. The easier way for me is to simply drop my offer and leave the bad stuff as it Jul 19 18:30:16 is. Jul 19 18:35:55 FINALLY Jul 19 18:36:00 I think I made it work Jul 19 18:44:01 need to tidy up its module loading, commit correct default config Jul 19 18:44:32 it segfaults if it tries to listen on an interface thats a member of a bridge but I'm not gonna even start fixing that, should just be configured properly Jul 19 18:47:31 mkresin: i realize i didn't address your concern about the 'crypto note': my rationale is simply to make it easier to find devices with this support if/when it's supported. If all mt7621 hardware has a crypto module then indeed it makes no sense to have the note and I'll remove it Jul 19 18:52:22 f00b4r0: one more thing about the fixed partitions thingy Jul 19 18:52:56 f00b4r0: using the node name will not work for all cases as you are limited in the char which can be used Jul 19 18:53:15 f00b4r0: the works I'm aware of is "tffs (1)". no way to express it as node name Jul 19 18:53:33 f00b4r0: hence you have to use a label for some and the nodename for some other Jul 19 18:54:13 f00b4r0: doesn't look like a big issue, but it makes the maintainance harder due "special cases" Jul 19 18:54:17 sure. That's exactly why I used it for the first partition as I was unsure it was ok to have caps in nodename. But why use it everywhere still? Jul 19 18:54:30 especially for common nodenames such as "firmware" Jul 19 18:54:36 f00b4r0: see one line above Jul 19 18:54:44 which should ultimately be found in all DTSes Jul 19 18:55:07 f00b4r0: it is a pain if something is done sometimes way a and sometimes way b Jul 19 18:55:31 i still don't see what could be the problem with having a property in some nodes and not in others. There are plenty of others similar cases, chiefly with GPIOs/LEDs Jul 19 18:55:32 f00b4r0: I'm considering running some kind of sed/awk/grep magic to add the fixed partitions parent node to all ramips dts Jul 19 18:55:49 f00b4r0: guess what will happen if they (unexpected) differ Jul 19 18:56:46 that's a shortcoming of the sed/awk/grep magic. Sounds to me like you want to use a dts-parser where you can manipulate the nodes themselves. The problem only exists if you try to treat the files as static text Jul 19 18:57:00 f00b4r0: as said, I'm not able to explain everything in a way that it is easy to understand for everybody. that are all lessons learned from taking care of a whole target or even the whole tree Jul 19 18:57:04 same as if you try to edit xml with sed/awk. It *might* work. It often won't. Jul 19 18:57:56 mkresin: what I read here "I want to make sure I can use my magic scripts on the files so I'm adding extraneous constraints to enforce that". Jul 19 18:58:11 forgive me if I find that to be a rather feeble argument ;P Jul 19 18:58:42 at that point you might define your own custom DTS subset to be used in owrt. Jul 19 18:58:56 f00b4r0: it is as simple as you may need to accept it as fact and need to trust in what I say. I'm doing this maintanance, PR review/merge since quite some time now. Jul 19 18:59:38 f00b4r0: and sure, I try to automate as much as I can to keep track with the number of PRs/patches Jul 19 18:59:39 ah. "trust what I say". That's another sure win argument for me. Might as well forego code review and trust the committers too. Jul 19 19:00:26 anyway, let's not argue again, we clearly see differently. I've made my points in the proposals, what happens next is what happens next. Jul 19 19:01:55 f00b4r0: mkresin: not sure if mentioned, there is some ongoing work by rmilecki to make these type of sub partions properly expressed in dts -> https://patchwork.ozlabs.org/patch/943613/ Jul 19 19:02:21 KanjiMonster: I stumbled upon that patch. Which kinda made me feel I wasn't doing anything wrong indeed. Jul 19 19:02:46 distemper: ping Jul 19 19:02:49 okay, I tried my best to explain why I'm deciding this or that way. but let us agree to disagree Jul 19 19:03:03 indeed. Jul 19 19:04:05 KanjiMonster: I wasn't aware of the patch. pretty much matches my proposal on the mailing list Jul 19 19:05:16 mkresin: i thought you objected to the very idea of top-level/overlapping partitions? Jul 19 19:05:29 f00b4r0: have you even read my reply Jul 19 19:05:34 i did Jul 19 19:05:43 f00b4r0: do it again Jul 19 19:06:30 quoting: "I neither see the need [...] to create the overlapping partitions [personal preference]" Jul 19 19:06:45 f00b4r0: keep on reading Jul 19 19:06:46 i must fail to parse english. Jul 19 19:07:17 well yes. You do realize rmilecki's patch hasn't been accepted yet? Jul 19 19:08:05 and forgive me if I focused on the beginning of your objections, considering what you said before both on irc and on github (where you _were_ objecting to overlapping partitions, period) Jul 19 19:09:14 f00b4r0: oh man, it is all about to express attempt to get something working which isn'T yet possible Jul 19 19:09:32 well, it will be if 943613 is accepted. Jul 19 19:09:54 but I doubt it'll get backported to 18? If so the current proposal should apply there. Jul 19 19:10:06 f00b4r0: by abue the bindings to not show a warning Jul 19 19:10:34 or is it already possible Jul 19 19:10:38 * f00b4r0 is confused now Jul 19 19:11:20 i'm not abusing anything Jul 19 19:11:37 the binding doesn't prevent overlapping partitions, and there's no reason why it should. Jul 19 19:11:58 f00b4r0: okay, I got the memo. I'm an idiot and you are the only one who knows how everything should be done Jul 19 19:12:03 after all, mbr 'extended' is one big overlapping partition... Jul 19 19:12:17 that's exactly what I said ;P Jul 19 19:14:01 mkresin: I thought that was psyborg? Jul 19 19:14:20 heh Jul 19 19:14:37 C'mon kids, play nice :) Jul 19 19:14:41 siding with him certainly didn't look good from my end ;P Jul 19 19:14:54 * f00b4r0 grabs a ball and plays nice Jul 19 19:15:00 is anyone using unprivileged lxc containers ? I am getting error after lxc-create https://pastebin.com/w0mbarRe Jul 19 19:17:02 so umm, with cmake.mk, what causes this sort of error? Jul 19 19:17:07 INSTALL FILES given unknown argument Jul 19 19:17:07 "/home/jwh/nspv2/staging_dir/toolchain-x86_64_gcc-7.3.0_musl/etc". Jul 19 19:17:15 can I fix it without patching sources? Jul 19 19:38:40 hm, is there no provision in git source downloads for the branch? Jul 19 19:39:09 so I can tell it to checkout something other than master Jul 19 19:48:38 Yo mean on packages? You can checkout an individual commit, but git will clone/download the repository anyway Jul 19 19:49:21 See https://openwrt.org/docs/guide-developer/packages Jul 19 19:49:21 search for "Complete git example" Jul 19 19:49:51 aaah Jul 19 19:50:12 doesn't look like I can say use branch X though instead of master Jul 19 19:50:18 nm Jul 19 19:50:33 can just generate a bunch of patches from commits for testing Jul 19 19:55:43 hm, wonder if I can save myself the hassle of reverting commits by using libtask Jul 19 20:24:26 hm maybe not, nothing directly emulating glibc Jul 19 20:24:37 guess I'll just have to start building my images with glibc instead :( Jul 19 20:35:58 at last serial console works Jul 19 20:36:17 o/ Jul 19 20:36:47 now the hard work begins Jul 19 20:43:34 is it possible to gather more info from uboot than just getting the boot log? Jul 19 21:20:13 jow: thanks for looking into and fixing the image builder problem Jul 19 22:33:10 is the lantiq ar platform supported? Jul 19 22:44:45 hex01: there are AR7, AR8 and AR9, see https://openwrt.org/docs/techref/hardware/soc/soc.lantiq for details - if your device is older than danube, I wouldn't bother though Jul 19 22:47:10 (you can find pretty nice VR9 based devices with plenty of RAM and flash for a fiver on the used market) Jul 19 23:02:49 pkgadd thanks, I'm just trying to find a device that's similar to mine but didn't see anything branded "AR10" Jul 19 23:15:35 hex01: if you're referring to https://forum.lede-project.org/t/support-for-tp-link-archer-d2-ver-1-0/17298/ I just answered there - AR10 is newer than AR9/ AR10 and doesn't have OpenWrt support yet, some parts should be similar enough to get working, others will need more work Jul 19 23:16:20 I just saw your post. I guess I'm way over my league then Jul 19 23:16:26 thanks for the info Jul 19 23:23:01 hm, I think I've managed to totally hose my tree Jul 19 23:23:07 * jwh rm -rfs Jul 19 23:32:31 jwh: a tip for speeding that up in future, make sure to always set the download directory to somewhere outside your buildroot, so it can be kept... Jul 19 23:33:45 ah good shout Jul 19 23:34:12 ok Jul 19 23:34:22 either I have some fucked up header/build_dir pollution Jul 19 23:34:26 or I'm going mad Jul 19 23:36:23 all this because musl suck Jul 19 23:36:24 :( Jul 19 23:37:42 posix compliance is all well and good, except the real world isn't compliant Jul 20 00:52:19 Hi there is a update to irqbalance Jul 20 00:52:44 latest is irqbalance release 1.3.0 Jul 20 00:53:01 openwrt is at irqbalance Jul 20 00:53:01 1.2.0-2 Jul 20 00:55:57 alright keep your hair on Jul 20 00:56:23 Hi jwh is that to me? Jul 20 00:56:58 yes :D Jul 20 00:57:11 why? I am balled. Jul 20 00:57:24 lol Jul 20 00:57:28 balled indeed Jul 20 00:57:30 * jwh chuckles Jul 20 00:57:34 :-D Jul 20 00:58:35 Any I was just putting it out there as I cant do the bump because I don't know what I am doing. Jul 20 00:58:59 i'm sure someone will get round to it Jul 20 00:59:03 if its worth bumping Jul 20 00:59:29 I was just going to say I don't even know how mutch use irqbalance is. Jul 20 01:00:29 I have just installed it but how do you start the thing? Jul 20 01:00:54 I cant find anything in the wiki Jul 20 01:01:39 fuck it's 2.01 here I had better get off to bed! Jul 20 01:01:46 its a method of balancing interupts between cores Jul 20 01:01:52 yup Jul 20 01:01:59 just finished watching a bit of The Killing Jul 20 01:02:01 Yeah I know what's for Jul 20 01:02:17 what it's for* Jul 20 01:02:42 After installing the package how do you start it? Jul 20 01:02:52 If you know. Jul 20 01:04:00 likely already running Jul 20 01:04:07 if it has an init script Jul 20 01:04:08 ps w Jul 20 01:04:10 K Jul 20 01:04:22 It does not say on the man page Jul 20 01:35:35 Have a question on building from source. I want to modify a few lines in project/uci. the Makefile refers to GIT, how can I download the sources for uci and build it with the fix? Jul 20 01:36:04 ok so it seems nginx doesnt generate a self signed cert if used as the default web server for luci ? Jul 20 01:36:37 stangri: cant you unpack it, and create a patch Jul 20 01:37:12 OutBackDingo: unpack what? I don't know how to create a patch, was hoping I could fix a few lines and that's it Jul 20 01:39:22 easy Jul 20 01:39:46 check out the commit in the makefile, make your fix, git diff > package/patches/blah.patch Jul 20 01:40:09 jwh: ty, let me try Jul 20 01:41:25 uhm, also, I can't find mvebu as a target in the snapshot :( Jul 20 01:41:27 you'll probably also want to apply any existing patches for that package Jul 20 01:41:31 what should I look for? Jul 20 01:42:08 uci isn't target speciic Jul 20 01:42:10 specific Jul 20 01:42:27 don't I need to pick the target for the binary? Jul 20 01:42:43 to build it yes, but not to patch it Jul 20 01:42:58 found it. Jul 20 01:46:15 git diff > package/patches/blah.patch -- package as in package folder or uci/patches/test.patch ? Jul 20 01:47:00 doesn't look like theres any patches (makes sense, its internally maintained) Jul 20 01:47:13 mkdir package/system/uci/patches, throw it in there Jul 20 01:47:13 uhm, sorry, I really need some hand holding thru this. Jul 20 01:47:19 ty Jul 20 01:49:43 jwh: should it print out any confirmation patch has been applied? Jul 20 01:50:04 yup, if you enable build output it will say applying patch Jul 20 01:50:09 may or may not see it if its quick Jul 20 01:50:15 (make -j1 V=s) Jul 20 01:50:33 I've just done this: make package/system/uci/{clean,compile} V=99 Jul 20 01:50:39 and it failed Jul 20 01:50:48 pastebin it Jul 20 01:51:24 https://pastebin.com/AC0CgcFp Jul 20 01:51:32 some dependencies are not compiled I guess Jul 20 01:52:04 have you done a build run first? Jul 20 01:52:06 or at least the toolchain Jul 20 01:52:11 nope Jul 20 01:52:16 just going by the wiki Jul 20 01:52:21 package on its own is pretty useless Jul 20 01:52:35 I have a live device I can test it on Jul 20 01:52:35 what page are you looking at? Jul 20 01:53:02 https://openwrt.org/docs/guide-developer/source-code/start and an old one: https://wiki.openwrt.org/doc/howto/build Jul 20 01:53:26 how can I build just the toolchain? Jul 20 01:53:37 it's an old computer, I don't want to run the full build Jul 20 01:57:21 umm Jul 20 01:57:27 its in the docs somewhere Jul 20 01:58:04 I never really tried doing just that (instead of doing a full build), but you at least have to start with some basic configuration as if you'd build a full image for your device, to get the config straight for the toolchain arch, the kernel arch, etc. Jul 20 01:59:30 toolchain kinda works Jul 20 01:59:47 dunno if the dependency chain will be complete though if trying to build just that package Jul 20 01:59:52 I'd really appreciate some help getting just basic stuff built Jul 20 02:00:03 also, if its a snapshot and you've cloned master, dependencies might be wrong anyway Jul 20 02:00:15 if I don't select any packages, just select the target platform and run make, would it build enough? Jul 20 02:00:53 for uci, probably yes - what mvebu device are you building for Jul 20 02:00:54 ? Jul 20 02:01:13 wrt3200acm Jul 20 02:01:27 can you try my patch and post the binary? Jul 20 02:02:20 this is the patch: https://pastebin.com/YyFdKwgE Jul 20 02:02:42 oh, that Jul 20 02:02:59 that http://paste.debian.net/1034449/ might be enough (testing, not finished yet) Jul 20 02:03:26 pkgadd: and then just make? Jul 20 02:03:45 store that as .config in the top level directory Jul 20 02:03:58 make -j32 defconfig oldconfig download Jul 20 02:04:00 make Jul 20 02:05:45 sat here, couldn't work out if I patched this wrong, or its just broken Jul 20 02:06:00 hypervisor doesn't send dhcp requests to vms Jul 20 02:06:02 * jwh cries Jul 20 02:06:49 its been a long (2) day(s) Jul 20 02:07:14 sleeping usually helps ;) Jul 20 02:07:15 grr, I should have built the toolchain before trying to apply the patch with quilt Jul 20 02:07:39 pkgadd: I'm glad I'm not the only on struggling here ;) Jul 20 02:07:44 not a problem, but -j1 make it slower than it needs to be Jul 20 02:07:52 heh, I've been pissing about with getting this package upstream worthy (and at least, working) for a couple of days Jul 20 02:08:06 between libc nonsense, broken code, dumbass cmake behaviour Jul 20 02:08:22 crap, should have checked out rc2... :( Jul 20 02:08:39 stangri: my biggest problem here is that I'm approaching it differently than I usually would, in order to save a little compilation time ;) Jul 20 02:08:41 if someone could just roll up getcontext and friends into libcontext or something Jul 20 02:08:44 that would be great Jul 20 02:09:36 pkgadd, jwh: thanks for the help, I'll let the make run for a while... hopefully I'll get an ipk or a binary out of this. Jul 20 02:10:03 you will, under bin/ Jul 20 02:10:59 pkgadd: get this for evil Jul 20 02:11:21 even the resulting firmware images could be flashed - but they're a bit bare bones (e.g. no luci) without extending the config a little Jul 20 02:11:39 package checks out an old revision, then applies a diff generated with git diff commit -- ':spec' to bring it up to master, sans a specific directory Jul 20 02:11:42 :DD Jul 20 02:12:20 there may be reasons for that (as in forward patching a release version) Jul 20 02:12:33 oh, no thats not why I did it Jul 20 02:12:53 I just don't have the patience to step through every single commit to figure out at what point they broke it Jul 20 02:13:13 need to check it actually works first Jul 20 02:14:29 at some point they switched to using ucontext stuff, which doesn't exist in musl Jul 20 02:15:00 but the API to the library hasn't changed, so it still works with an older version Jul 20 02:15:30 asshole move defining the functions in headers though, on musls part Jul 20 02:22:13 pkgadd: https://github.com/joeholden/staging-packages/tree/master/accel-ppp Jul 20 02:22:16 pretty grim Jul 20 02:38:56 jwh: still woke? ;) can I comment in some way to approve/confirm this: https://github.com/openwrt/luci/pull/1967 to get it moving along? Jul 20 02:39:51 nope Jul 20 02:39:54 if you have a github account, you can comment on the pull request Jul 20 02:40:15 it'll be merged when its merged :D Jul 20 02:48:06 jwh: I meant do you normally need a second opinion from someone who has command of the language to get it merged? Jul 20 02:48:38 like what should I say to confirm it's valid? Jul 20 02:49:38 no idea about translations Jul 20 02:49:57 I'd imagine an ACK from a native speaker would do **** ENDING LOGGING AT Fri Jul 20 03:00:01 2018