**** BEGIN LOGGING AT Sat Nov 28 02:59:56 2020
Nov 28 03:26:00 <dorf> just installed latest snapshot. installing the openwrt2020 luci theme appears to kill luci
Nov 28 03:26:33 <dorf> running snapshot in a vm, fwiw.
Nov 28 03:27:11 <pkgadd> I've (successfully) used that (self-built) recently (as in r15064-66732c5cd9, ~two days ago)
Nov 28 03:28:04 <dorf> that specific theme?
Nov 28 03:28:12 <pkgadd> perhaps
Nov 28 03:28:21 <dorf> default theme works a.ok
Nov 28 03:28:27 <pkgadd> https://github.com/openwrt/luci/commit/03c77dafe3cfb922b995adfe9c0f8a75c98a18af 'might' be a potential cause
Nov 28 03:28:48 <dorf> is that the reversion from divs to tables jow committed earlier?
Nov 28 03:28:55 <pkgadd> yes
Nov 28 03:29:09 <dorf> could be that, sure.
Nov 28 03:29:14 <pkgadd> not saying that this *is* the issue, I simply don't know
Nov 28 03:29:28 <dorf> not critical, just thought I'd mention it.
Nov 28 03:29:54 <dorf> in other news, my patched bootstrap theme isn't showing any issues in snapshot vs 19.04
Nov 28 03:31:50 <dorf> the issues jow raised earlier I'll be working on presently.
Nov 28 03:32:45 <dorf> and I noticed that snanshot now has an adguard package.. are there any plans to create a luci-app for that?
Nov 28 03:33:26 <dorf> not that the existing adguard ui is bad, but would be nice to be able to manage it from luci.
Nov 28 03:51:52 <mangix> aparcar[m]: wasn't the omnia PR two commits?
Nov 28 03:52:25 <aparcar[m]> yes but rebasing drop the file
Nov 28 03:52:31 <aparcar[m]> so not it's three commits
Nov 28 03:52:34 <mangix> oh ok
Nov 28 04:04:39 <philipp64> mangix: definitely looks like python3 is failing to build, but oddly it’s not even selected. http://paste.ubuntu.com/p/5TsG4PQ87v/
Nov 28 04:06:24 <mangix> that's still incomplete
Nov 28 04:06:44 <aparcar[m]> mangix: any burning patches to look at?
Nov 28 04:06:45 <mangix> run make package/python/clean && make package/python/compile V=s
Nov 28 04:07:34 <mangix> aparcar[m]: https://patchwork.ozlabs.org/project/openwrt/patch/20200725012009.223148-1-rosenp@gmail.com/
Nov 28 04:07:37 <mangix> simple one
Nov 28 04:13:35 <mangix> what was the concensus on iftop? move to packages?
Nov 28 04:16:16 <aparcar[m]> I'm not the maintainer
Nov 28 04:16:55 <aparcar[m]> how do I test the gettext-full one again?
Nov 28 04:17:04 <aparcar[m]> as in, I want to ee it break and then I want to see it work
Nov 28 04:17:47 <mangix> make package/glib2/host/compile && make package/gettext-full/host/clean && make package/gettext-full/compile
Nov 28 04:20:31 <aparcar[m]> ok it needs a bit to compile
Nov 28 04:20:42 <mangix> yeah it's a garbage package
Nov 28 04:20:51 <mangix> doesn't build in parallel
Nov 28 04:20:57 <mangix> and configure takes forever
Nov 28 04:22:12 <aparcar[m]> meh
Nov 28 04:22:17 <aparcar[m]> isn't there something better?
Nov 28 04:22:25 <mangix> than what?
Nov 28 04:22:30 <aparcar[m]> gettext ;)
Nov 28 04:23:02 <mangix> I...have no idea
Nov 28 04:23:36 <mangix> there's a newer version, but I haven't updated it as I can't get it to compile
Nov 28 04:24:01 <aparcar[m]> mangix: btw is anyone working on btrfs?
Nov 28 04:24:09 <mangix> not that I know of
Nov 28 04:24:43 <mangix> that subvolume proposal is extremely interesting though
Nov 28 04:25:19 <aparcar[m]> yes
Nov 28 04:25:23 <aparcar[m]> but I didn't see anyone working on that
Nov 28 04:25:54 <mangix> right. it was just an idea
Nov 28 04:32:04 <aparcar[m]> woa still combiling...
Nov 28 04:32:26 <mangix> i told you
Nov 28 04:33:18 <mangix> FFS
Nov 28 04:33:26 <mangix> there's another piece to getting it to fail
Nov 28 04:33:32 <mangix> you need to build with full NLS
Nov 28 04:42:21 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_omap.html has been updated. (0% images and 97.4% packages reproducible in our current test framework.)
Nov 28 04:44:40 <aparcar[m]> mangix: did you test https://patchwork.ozlabs.org/project/openwrt/patch/20201127220429.22458-1-ynezz@true.cz/?
Nov 28 04:45:00 <mangix> kind of sort of
Nov 28 04:45:21 <mangix> it probably works, but ccache fails in many other places (unrelated to that patch)
Nov 28 04:45:43 <aparcar[m]> mh
Nov 28 04:45:49 <aparcar[m]> the indention looks bad
Nov 28 04:46:41 <mangix> anyway, the original bug report was about tools/ failing to compile. with that patch, they compile
Nov 28 04:46:48 <aparcar[m]> mangix: I don't get it to fail, sorry
Nov 28 04:46:53 <aparcar[m]> the gettext thing
Nov 28 04:47:01 <mangix> it's not that compilation fails
Nov 28 04:47:12 <mangix> it's that the linked libraries are wrong
Nov 28 04:48:02 <mangix> let me see if I can get it to fail
Nov 28 04:51:43 <philipp64> mangix: no joy. can you have a look?
Nov 28 04:52:11 <mangix> philipp64: python is not failing on my end
Nov 28 04:53:16 <philipp64> is it ‘python’ or ‘python3’?
Nov 28 04:53:46 <mangix> ah it is python3
Nov 28 04:53:54 <mangix> i thought they renamed it
Nov 28 04:54:21 <philipp64> and to build the host version?
Nov 28 04:55:55 <mangix> make package/python3/host/compile
Nov 28 04:59:05 <mangix> huh weird
Nov 28 04:59:15 <mangix> I can't reproduce that failure
Nov 28 04:59:37 <mangix> w/e. I don't feel like finding out why
Nov 28 05:00:23 <mangix> wait...i have an idea
Nov 28 05:09:04 <philipp64> did the clean then a -j1 build… http://paste.ubuntu.com/p/SKKxhvFYCx/
Nov 28 05:10:34 <mangix> find build -type f -a ! -name '*.gc??' -exec rm -f {} ';'
Nov 28 05:10:34 <mangix> find: 'build': No such file or directory
Nov 28 05:10:47 <mangix> i have no idea what that is
Nov 28 05:13:02 <philipp64> anything get updated recently that might have broken the build?
Nov 28 05:15:17 <mangix> besides the version bump, no
Nov 28 05:18:00 <philipp64> which one? looks like there were several.
Nov 28 05:20:07 <mangix> the 3.9.0 one
Nov 28 05:20:26 <mangix> aparcar[m]: I can't replicate the failure.
Nov 28 05:24:01 <aparcar[m]> mangix: Sounds good
Nov 28 05:24:53 <mangix> it's probably the configure script failing. i don't care enough for the package to figure it out
Nov 28 05:54:08 <mangix> aparcar[m]: uhhhhh why are there two audit packages
Nov 28 05:54:25 <mangix> they both use the same tarball...
Nov 28 06:29:21 <dorf> jow: if you're around and about, I've addressed the issues you raised in my latest commit, all except the bold font issue (install open-sans-ttf!)
Nov 28 11:09:19 <russell--> what are workingset_refault in /proc/vmstat?
Nov 28 11:13:26 <russell--> adding eth0 (on a ubiquiti bullet m5) makes my memory usage explode (from about 12MB to 20MB) before it just helpfully rests for me
Nov 28 11:13:37 <russell--> adding to a bridge*
Nov 28 11:14:07 <russell--> (batman-adv bat0 interface and a wlan0 are the other members of the bridge)
Nov 28 11:22:13 <russell--> something to do with cgroup2?
Nov 28 11:23:55 <russell--> reboot-15093-g501123eb7a
Nov 28 12:06:53 <russell--> disabling cgroups and namespaces seems to help
Nov 28 12:09:58 <russell--> only temporarily
Nov 28 12:10:03 <russell--> about to oom again
Nov 28 12:13:16 <russell--> oom killed netifd
Nov 28 12:13:48 <russell--> and my memory came back
Nov 28 12:27:29 <russell--> whether ethernet is in the bridge or not seems to be mostly irrelevant, just having it plugged in causes leakage, killing netifd (and it restarting) seems to recover memory
Nov 28 12:40:37 <russell--> top --help
Nov 28 12:59:54 <russell--> if i take the wifi interfaces down (wifi down), then plugging the eth0 in doesn't seem to cause the leakage
Nov 28 13:20:56 <adrianschmutzler> mangix: you can add the four libnetfilter packages to package repository
Nov 28 13:21:22 <adrianschmutzler> any reason why you did not do the same for libnetfilter-conntrack?
Nov 28 13:21:41 <adrianschmutzler> apart from that we should ask jow if that seems reasonable
Nov 28 13:44:15 <mangix> it's used in the base feed
Nov 28 13:44:40 <adrianschmutzler> okay, didn't find it via grep
Nov 28 13:44:45 <mangix> dnsmasq and iproute2 if I remember correctly
Nov 28 13:46:12 <adrianschmutzler> ah, fact, didn't look carefully enough
Nov 28 13:47:09 <adrianschmutzler> but the other four seem fine
Nov 28 13:47:18 <adrianschmutzler> will push to master as soon as they landed in packages
Nov 28 13:47:26 <adrianschmutzler> btw, 3 of 4 don't have a maintainer
Nov 28 13:49:04 <mangix> merged
Nov 28 13:50:24 <adrianschmutzler> merged, too.
Nov 28 13:50:28 <adrianschmutzler> thanks for the quick response.
Nov 28 13:52:12 <russell--> batman doesn't seem to be implicated ... i removed the batman-adv config and still see the leakage
Nov 28 14:03:49 <russell--> even in dumb-ap mode, leakage
Nov 28 14:16:27 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_lantiq.html has been updated. (96.1% images and 92.0% packages reproducible in our current test framework.)
Nov 28 14:55:46 <ynezz> russell--: are you running odhcp6c?
Nov 28 15:15:54 <owrt-snap-builds> build #707 of at91/sam9x is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/at91%2Fsam9x/builds/707
Nov 28 15:25:37 <damex> uh... why do we need kernel2minor on mikrotik devices?
Nov 28 15:38:51 <owrt-snap-builds> build #458 of bcm47xx/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm47xx%2Fgeneric/builds/458
Nov 28 15:49:23 <owrt-snap-builds> build #736 of sunxi/cortexa53 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/sunxi%2Fcortexa53/builds/736
Nov 28 15:58:20 <owrt-snap-builds> build #733 of ipq806x/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/ipq806x%2Fgeneric/builds/733
Nov 28 16:07:20 <owrt-snap-builds> build #714 of arc770/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/arc770%2Fgeneric/builds/714
Nov 28 16:25:02 <f00b4r0> dorf: just my humble opinion here but it seems counter-intuitive to expect (let alone require) users to install a specific font to be able to get the desired theme experience. Can't you make this work with standard system fonts?
Nov 28 16:29:17 <owrt-snap-builds> build #533 of ath79/tiny is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/ath79%2Ftiny/builds/533
Nov 28 17:00:06 <owrt-snap-builds> build #533 of sunxi/cortexa8 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/sunxi%2Fcortexa8/builds/533
Nov 28 17:02:26 <owrt-snap-builds> build #583 of x86/legacy is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Flegacy/builds/583
Nov 28 17:17:10 <owrt-snap-builds> build #548 of mxs/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/mxs%2Fgeneric/builds/548
Nov 28 17:18:52 <owrt-snap-builds> build #528 of ipq40xx/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/ipq40xx%2Fgeneric/builds/528
Nov 28 17:55:03 <owrt-snap-builds> build #616 of mvebu/cortexa9 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/mvebu%2Fcortexa9/builds/616
Nov 28 18:10:23 <owrt-snap-builds> build #564 of octeon/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/octeon%2Fgeneric/builds/564
Nov 28 18:14:46 <philipp64> mangix: rolling back to pre-3.9.0 and Python still won’t build.
Nov 28 18:16:56 <philipp64> It’s not selected… nothing requires it… but it’s trying to build anyway.
Nov 28 18:30:17 <owrt-snap-builds> build #347 of ath79/mikrotik is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/ath79%2Fmikrotik/builds/347
Nov 28 18:52:03 <owrt-snap-builds> build #615 of x86/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Fgeneric/builds/615
Nov 28 19:02:07 <owrt-snap-builds> build #397 of bcm47xx/mips74k is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm47xx%2Fmips74k/builds/397
Nov 28 19:29:49 <johnf> I'm interested in this bug affecting 19.07.4 https://bugs.openwrt.org/index.php?do=details&task_id=3457
Nov 28 19:29:58 <johnf> are there patches staged for 19.07.5 ?
Nov 28 19:30:34 <johnf> the current snapshot, vs 19.07.4, contains extremely large amounts of changes
Nov 28 19:31:05 <johnf> so if I want to try to fix the issue there, should I start from 19.07.4 or is there some other way?
Nov 28 19:34:47 <dorf> f00b4r0: it's all about the font weights and what default fonts an OS supplies. there is no one-size fits all font that supports multiple weights, unfortunately, and the binary bold/normal choice inherent with most fonts isn't sufficient.
Nov 28 19:36:23 <f00b4r0> dorf: my understanding is that this can be dealt with in CSS by specifying a few "well known fonts"? Point being: I don't see a theme that requires a font install having great user popularity.
Nov 28 19:36:26 <f00b4r0> but maybe I'm wrong.
Nov 28 19:36:49 <f00b4r0> there's the alternative to use Google Fonts, but that requires connectivity and may also be frowned upon.
Nov 28 19:37:18 <dorf> a few well-known fonts are already specified. there's probably some more work to be done there to indentify os-resident fonts, sure.
Nov 28 19:37:50 <dorf> the other alternative is to supply the fonts with the theme.
Nov 28 19:38:32 <f00b4r0> I'm also thinking that installing extra fonts isn't an option on mobile.
Nov 28 19:39:25 <dorf> adguard, which is now shipping as an optional package in snapshot builds, has the same issue, fwiw.
Nov 28 19:40:51 <dorf> I think maybe supplying the fonts with the theme and placing them in a location such that they can be used globally _might_ be the way to go, for a uniform experience regardless of os/platform.
Nov 28 19:41:49 <dorf> I mean, we're no longer dealing with 4MB flash storage, so the size of the fonts shouldn't present a major issue.
Nov 28 19:46:19 <dorf> Could even be a UI option under theme selection that allows for manual downloading of the required font family if a suitable one isn't present on the client accessing LuCI.
Nov 28 19:52:23 <dorf> f00b4r0: the question really is, though, have you tested the theme (with open sanss) and do you see the benefit of multi-weight fonts? :)
Nov 28 19:56:27 <dorf> jow: the realtime connections table could do with a look.. the javascript is regularly causing firefox to suggest stopping the running scripts on the page.
Nov 28 19:56:43 <dorf> > /cgi-bin/luci/admin/status/realtime/connections
Nov 28 20:09:09 <dorf> maybe lazy-loading the table contents might be an idea.
Nov 28 20:16:42 <f00b4r0> dorf: I'm fully aware of the merits of multi-weight fonts, all I'm saying is: outside of a few power users, no one is going to install fonts (assuming they can, which is not always the case) just to enjoy a router UI theme ;-)
Nov 28 20:16:59 <f00b4r0> as for storage size, 4MB are gone, but 8MB are still very much with us...
Nov 28 20:18:27 <f00b4r0> and based on some websites I've seen working out splendid (and consistent across platforms) UI with system fonts (Segoe, San Francisco, etc), I'd innocently thought it'd be possible to do the same :)
Nov 28 20:18:54 <dorf> sure, you're referencing OS X and Windows fonts.
Nov 28 20:19:06 <dorf> Segoe UI is already provisioned.
Nov 28 20:19:06 <f00b4r0> I don't remember the Android one
Nov 28 20:20:02 <dorf> as is Ubuntu Sans. Linux/Android/OS X, work to do. Happy to take recommendations.
Nov 28 20:33:23 <f00b4r0> dorf: github uses "-apple-system". Looks good to me ;-)
Nov 28 20:33:49 <f00b4r0> renders as San Francisco as far as I can tell
Nov 28 20:34:23 <dorf> ok, that's possibly dependent on which variant of OS X you're running. I can hook that in no problemo. Adguard is also using -apple-system
Nov 28 20:34:58 <f00b4r0> makes the UI feel very much "at home"
Nov 28 20:35:27 <dorf> sure, that's really the intention. done. let me push that. one moment please...
Nov 28 20:35:44 <f00b4r0> cool!
Nov 28 20:35:48 <f00b4r0> have to bail now, bbl
Nov 28 20:37:11 <dorf> o/
Nov 28 20:43:28 <dorf> done, f00b4r0, for when you return.
Nov 28 21:12:49 <owrt-snap-builds> build #635 of at91/sama5 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/at91%2Fsama5/builds/635
Nov 28 22:53:38 <russell--> ynezz: no
Nov 28 22:54:40 <russell--> this is configured as a bridge, no iptables at all, no dhcp at all.
Nov 28 22:56:31 <russell--> i am going to try to trim the config down and see if i can bisect it. pretty sure was not seeing this as of last spring
Nov 29 02:34:48 <russell--> i get some libpam warnings when building without feeds
**** ENDING LOGGING AT Sun Nov 29 02:59:57 2020