**** BEGIN LOGGING AT Sun Oct 14 03:00:00 2018 Oct 14 10:53:25 jow: ping? Oct 14 11:17:55 jow: I compile tested all the hostapd variants, are you fine with me merging this Oct 14 11:18:51 dedeckeh: nbd: how do you create the commit messages for the netifd, procd, .. updates in OpenWrt? Oct 14 11:20:59 Hauke: git log --oneline --reverse --no-decorate Oct 14 11:30:42 jow: /me doesn't understand the rollback thing in luci these days, today had a change to the lan ipaddr rollback repeatedly, wasn't sure what we were supposed to do to complete/confirm the change (not a habitual luci user). Oct 14 11:37:45 russell--: when it times out after 30 sec and says 'I've rolled back' you click the 'apply these changes anyway' Oct 14 11:38:46 what is has done (for an ip change at least) .... it changes the ip address of the device and waits for the browser to say 'hi I'm here'. Oct 14 11:39:05 the browser is talking to the old address.... the device is listening on the new address. Oct 14 11:39:45 ldir: how does the "apply these changes anyway" message get to it if the config has changed? Oct 14 11:40:02 or, that's already after the config rolled back? Oct 14 11:40:09 the browser & device don't get any response, so the device rolls back...to the old address.... and the browser now talks and says 'that failed so the device rolled back'.... but if you click 'apply changes anyway' then it re-applies the settings again. Oct 14 11:40:42 if you are too slow, you end up chasing your tail Oct 14 11:41:02 and then the 'really' fun bit.... the browser still tries to talk using the old IP address...which will fail (again) and then comes back and says 'your device is offline'. - it isn't....it's just listening on the new address :-) Oct 14 11:41:59 * russell-- just turned off the browser, and did the change with serial console and vi Oct 14 11:42:00 There are two scenarios discussed here recently that are 'challenging' with the rollback behaviour... IP address change and PSK change if accessing device over wifi. Oct 14 11:42:39 nbd: thanks Oct 14 11:44:06 the IP address change via browser isn't a case of being quick enough...it's really a case of the user waiting 30 seconds for the first 'it failed prompt' (which it didn't) and then saying 'go do it anyway'. and *then* pointing the browser at the new address. Oct 14 11:45:34 * russell-- isn't convinced this is a win ... eventually you figure out you need to reconfigure and chase the thing that moved, but if the thing that moved is still moving, super confusion Oct 14 11:46:23 it's funny you say that 'cos jow is thinking of another technique..... that I'm not convinced is progress either. Oct 14 11:48:09 * russell-- imagines washing face with soap in shower, puts soap in dish with eyes closed, then reaches back to find soap but it isn't there anymore because it helpfully moved somewhere else Oct 14 11:48:18 and to be fair, the existing behaviour has actually saved me a few times...... it's just not so helpful/intuitive with IP changes. Oct 14 11:48:52 russell--: ah, wife has helpfully moved soap to make things tidy! ;-) Oct 14 11:56:22 how does mixture of luci and shell-originated uci commands interact? Oct 14 11:58:01 we made a change from luci, it reverted after a timeout, tried to make a uci set network.lan.ipaddr=foo ; uci commit network, but that seemed to get rolled back too. Oct 14 12:03:46 no idea Oct 14 13:47:01 Hauke: thanks for sorting out the patch v automake thing Oct 14 13:53:29 ldir: looks like it didn't fix the problem Oct 14 13:58:06 oh :-( is this the tools/patch patch or the packages/patch patch ? Oct 14 13:58:49 is it revert time? Oct 14 13:59:33 ldir: let me have a look it complains about Circular dependecies Oct 14 14:46:21 well that's,,,unexpected Oct 14 14:47:15 I call for the removal of building tests for patch since that's where the problem seems to be. Oct 14 14:49:47 funny you should mention that, I have just this second removed that part of the patch which adds ed-style to the tests/Makefile.am Oct 14 14:50:27 ldir: now it should work again Oct 14 14:50:38 I feel like an ignorant vandal.....but it does appear to work. Oct 14 14:50:59 * ldir goes to look at the proper solution..... git fetch Oct 14 14:51:04 ldir: tools/makefile adds a dependecy to patch to all tools which have a patches folder Oct 14 14:51:24 so removing the patch is the easiest fix, then it does not run automake again Oct 14 14:53:02 hah! so you've actually done exactly what I just did when I thought I was being an ignorant vandal :-) Oct 14 14:54:02 ldir: yes, this was more of a chicken or the egg problem ;-) Oct 14 14:54:49 lol - well it's the chicken clearly, else from where do you get the egg ? ;-) Oct 14 14:59:25 ldir: looks like it is building again: http://phase1.builds.lede-project.org/builders/ar71xx%2Fmikrotik/builds/1047/steps/tools/logs/stdio Oct 14 15:02:32 yay - as is my own system (after replicating the error with a dirclean) Oct 14 15:05:08 WTH Samsung!? So you have an option in Settings where you can choose from a number of pre-defined names for your TV *BUT* even after doing so, you *STILL* try to get a DHCP lease for hostname 'localhost' Oct 14 16:51:06 ldir: don't connect your tv to the internet, it usually makes stuff worse ;) Oct 14 16:51:26 also just give it a static lease Oct 14 16:59:58 I think the relevant devs at Samsung just need the application of a Cluebat. Oct 14 19:27:41 Hauke: thanks for sorting that out Oct 14 19:29:55 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Oct 14 19:51:52 nbd: ping Oct 14 20:10:13 i had a senior moment and cant remember the git command to update my local openwrt repo.. Oct 14 20:44:32 git pull --ff-only? Oct 15 01:16:42 man its been a while Oct 15 01:16:53 theres no way to do a graceful sysupgrade on x86 right? Oct 15 01:17:31 wonder if its doable to add the same sort of sysupgrade proces that other targets have Oct 15 01:26:39 you can sysupgrade just fine as long as it will fit on the boot partition Oct 15 01:29:50 well I mean, I could manually dd it, yeah Oct 15 01:30:07 but being able to do sysupgrade would be nice, especially if it can restore configs Oct 15 01:30:16 you can Oct 15 01:30:34 i do it every time Oct 15 01:30:48 what do you give sysupgrade as the image? Oct 15 01:30:54 i even did it onto different hardware Oct 15 01:31:04 combined img gz Oct 15 01:31:10 oh Oct 15 01:31:14 didn't think it understood that Oct 15 01:31:50 works great with squashfs imgs Oct 15 01:32:23 hm, would certainly make my life easier Oct 15 01:33:34 oh thats cute Oct 15 01:33:36 thanks Oct 15 01:34:11 unannounced maintenance again this week, gonna get stabbed Oct 15 01:34:12 :D Oct 15 01:35:16 holy shit it even booted first time :D Oct 15 01:35:47 except obviously I'm missing the nic drivers D Oct 15 01:35:48 :D **** ENDING LOGGING AT Mon Oct 15 03:00:01 2018