**** BEGIN LOGGING AT Mon Dec 16 03:01:54 2019 **** BEGIN LOGGING AT Mon Dec 16 07:26:24 2019 Dec 16 07:30:50 mangix: it does Dec 16 07:31:20 but all files in these directories will be backed up unconditonally, so no change tracking Dec 16 09:02:41 hexa-: nxdomain afaik Dec 16 11:23:04 any idea how send anything to ubus from the shell (writing directly to unix socket in /var/run/ubus.sock) in the default OpenWrt install (not installing socat or so) ? Dec 16 11:24:52 ubus call foo bar '{ "data": "blah" }' ? Dec 16 11:26:45 ah no directly... not sure. Woul you even be able to properly format the blobmessages using shell only? Dec 16 11:27:32 anyhow, I guess the short answer is "no" Dec 16 11:29:57 right, I need to bypass that json2blobmsg translation, so writing directly to the socket Dec 16 11:32:07 what's the motivation here for my curiosity? Dec 16 11:38:49 just my curiosity Dec 16 11:40:23 dont have that intimate knowledge about unix sockets and found it quite weird that simply echo > socket file doesn't work Dec 16 11:41:37 echo -en '\x00\x04\x01\x00\x00\x00\x00\x00' > /var/run/ubus.sock Dec 16 12:04:45 ynezz: sockets require a connect call prior to write Dec 16 12:12:34 karlp: fuzzing and/ore testing malformed payloads would be one main motivation Dec 16 12:13:00 s/main/possible/ Dec 16 12:13:14 yeah, but then you'd be ok with installing extra tools I would have thought Dec 16 12:13:34 depends Dec 16 12:13:49 hrmm, sure, guess depends what you're "reserching" :) Dec 16 12:14:09 I coulkd imagine that requiring feed packages for an automatic ci setup could be problematic Dec 16 12:23:53 blogic: yep, but I'm already telling the destination by "connecting" to that file, so why the extra step Dec 16 12:24:47 karlp: imagine you've airgaped device Dec 16 12:29:50 I would like to have some convenient way, which one could just copy&paste into shell in order to verify it, so the less dependencies, the better Dec 16 13:24:12 anyone tried acme.sh in 19.XX ? seems its not working, at least DNS key=val does not work ( it does not write keys,values to account.conf ) Dec 16 14:34:54 why is libsmartcols so big?! 77k for colums and tables? Dec 16 14:35:24 because is smart? Dec 16 14:35:27 scnr :) Dec 16 14:36:38 scnr? Dec 16 14:36:48 sorry could not resist Dec 16 14:40:20 np :) not like I can really do anything with smartcols anyway, used all over util-linux. Dec 16 15:17:30 blogic: Can you have a look at the following two device support PRs which are interfering with old ramips_set_preinit_iface()? I'm just interested in a verdict on the changes there, no need to review the whole PR (as you seem to be the main author of this script...) Dec 16 15:17:36 https://github.com/openwrt/openwrt/pull/2460/commits/66f83cc5d0d170a621be9ccb2222421f985b50d2 Dec 16 15:17:51 https://github.com/openwrt/openwrt/pull/2540/commits/d30272e22b20e35e157b48615f21972e1923890a Dec 16 15:26:38 a verdict on what? how are they interfering? is there anything wrong with it now that they added ac omment? Dec 16 15:42:30 adrianschmutzler: looks legit Dec 16 15:42:44 adrianschmutzler: mt7621 has a builtin 100mbit mt7530 switch Dec 16 15:42:50 but some unit also have an external one Dec 16 15:43:00 the kernel will internally figure this out via dts Dec 16 15:43:09 but during failsafe that hack looks correct Dec 16 15:44:53 okay, and it's also necessary to use another selection mechanism there (board name instead of this RT3X5X?) Dec 16 15:45:01 ? Dec 16 15:45:44 yes, mt7530 is also available as separate chip Dec 16 15:46:10 netis is not mt7621, or is it? Dec 16 15:46:23 netis,wf2770 Dec 16 15:46:37 SoC: MT7620A Dec 16 15:48:22 so on netis its two devices, mt7620a soc + mt7530 switch and on mt7621 its single soc (has mt7530 switch IP block inside) Dec 16 15:48:58 ah, now I get it Dec 16 15:49:10 btw instead of those noisy comments, it might be better to just use properly named functions Dec 16 15:49:18 (my bikeshed of the day :)) Dec 16 15:50:13 Well, my understanding of that part is too limited to tidy it up ... Dec 16 15:52:05 but thanks for looking at the issue, I have to leave now. Dec 16 16:00:08 adrianschmutzler: it just looks like 1st case gets the switchdev_from_soc, 2nd case gets the switchdev_from_board, where switchdev_from_board should've priority (switchdev = $(switchdev_by_board || switchdev_by_soc)) and would as well save one unnecesary grep invocation Dec 16 17:10:32 ynezz: something like that: https://git.openwrt.org/?p=openwrt/staging/adrian.git;a=commitdiff;h=38f1fd4f2688d8cf98c9d0c683f358b4494a35c3 Dec 16 17:11:17 can I call functions in functions when using boot_hook_add? Dec 16 17:13:27 should be no problem Dec 16 17:23:30 adrianschmutzler: great, thanks! Dec 16 17:26:21 adrianschmutzler:you still have a -n ralink_switchdev there, the old var name? Dec 16 17:26:53 karlp: Yes, I've updated with some other small changes: https://git.openwrt.org/?p=openwrt/staging/adrian.git;a=shortlog;h=refs/heads/staging Dec 16 17:27:24 But thanks for spotting Dec 16 17:36:11 Okay, thanks for the feedback, I've written a comment to mans0n, so he can consider using/testing the changes now. Dec 16 17:39:02 Despite, I'm not familiar with the setup of those switches in DTS, too, so if someone just had a look at the ethernet node for the A1004ns, I'd take care of the rest. Dec 16 17:39:14 too->either Dec 16 17:40:08 https://github.com/openwrt/openwrt/pull/2460/commits/66f83cc5d0d170a621be9ccb2222421f985b50d2 Dec 16 18:10:06 jow: don't particularly care about version tracking. The backup directory has files like UUID.plist **** BEGIN LOGGING AT Mon Dec 16 22:39:25 2019 **** ENDING LOGGING AT Tue Dec 17 02:59:58 2019