**** BEGIN LOGGING AT Tue Jan 28 02:59:57 2020 Jan 28 07:35:24 Hauke: done Jan 28 07:43:51 rmilecki: ping Jan 28 09:34:29 jow: is there anything "wrong" with just accessing the L object from any luci rendered view? I've got some classic lua template htm+js files, and it appears I can just access the L object and make calls, and it's awesome, and will let me toss out heaps of old code, both controller side and js side, are there going to be any traps? Jan 28 09:41:20 karlp: accessing L methods (http://openwrt.github.io/luci/jsapi/LuCI.html) directly is fine Jan 28 09:41:59 karlp: you should not access classes through it though. e.g. L.ui.whatever() is prone to race conditions while 'require ui'; [...]; ui Jan 28 09:42:09 .. ui.whatever(); is safe Jan 28 09:46:26 I was mostly looking at things like L.uci and .ubus and .rpc Jan 28 09:47:07 these are loaded after luci.js so they might be prone to races Jan 28 09:48:28 what will they race with? how will this manifest? Jan 28 09:48:56 to be safe, you should use Promise.all([L.require('uci'), L.require('rpc')]).then(function(res) { var uci = data[0], rpc = data[1]; /* actual code */ }) Jan 28 09:49:30 it might lead to spurious null accesses when the cache is populated or the connection very fast Jan 28 09:49:48 e.g L.uci is null yet while your code already runs Jan 28 09:50:02 because the class is still loading Jan 28 09:50:13 it really depends on the context though Jan 28 09:51:11 hrm, this worked nicely on the console yesterday, and it's not today. I'll poke a bit more. I've got blobs of cruft from years ago that I think can mostly be removed though. Jan 28 09:51:41 are you using a server side Lua template with JS or a pure client side view? Jan 28 09:51:52 server side lua template with js. Jan 28 09:52:21 in that case I suggest wrapping the code in the Promise.all construct written above Jan 28 10:02:34 L.url is ~== luci.dispatcher.build_url(), and L.resource is ~== <%=resource%>, but is there a method for <%=media%> ? Jan 28 10:07:25 no, not yet. Jan 28 10:10:47 jow: could i ask you nicely to see qery in #openwrt ? Jan 28 10:15:33 karlp: L.media() pushed Jan 28 10:19:26 groovy thanks :) Jan 28 10:25:21 re races and promises, is simply checking for the existing of L. not enough? Jan 28 10:25:32 (thought taht may be worse style from a js developer pov) Jan 28 10:38:56 karlp: sure woudl work too, but what then? Simply don't run your code in 1 of 100 cases? Jan 28 10:39:56 or is that in an interval loop? Then it shouldn't matter Jan 28 14:39:41 nbd: are you fine with the backport of `kernel: fix dst reference leak in flow offload` into 19.07? https://github.com/openwrt/openwrt/pull/2721 Jan 28 14:40:58 yes, though probably the commit msg should be adjusted to match the regular backport format Jan 28 14:41:09 sure, I'll do that Jan 28 14:41:11 thanks Jan 28 15:10:21 [florian]: neoraider: mirko: could you cast your vote anytime soon http://lists.infradead.org/pipermail/openwrt-adm/2020-January/001253.html ? :-) Jan 28 15:11:33 Noltari: ^ Jan 28 17:33:30 blogic, you around? Jan 28 23:12:14 ynezz: sorry, took care of that. I tend to ignore all mailing lists when I'm busy ... Jan 28 23:21:01 neoraider: thanks for checking IRC :) Jan 29 01:04:07 <[mbm]> can someone explain the current state of wifi isolation to me? appears as though ap_isolate is forced on in the hostapd config, with the intention that the bridge then does the de-isolation? Jan 29 01:04:37 <[mbm]> (doesn't seem to be working, so I'm trying to debug how to disable isolation) Jan 29 01:12:59 [mbm]: what version? Jan 29 01:13:28 <[mbm]> 19.07.0 Jan 29 01:19:22 <[mbm]> russell--: ^ Jan 29 02:06:49 <[mbm]> hmm .. arp requests are failing wlan<->wlan but if I force the arp I can push data through Jan 29 02:07:09 <[mbm]> which means I need to start looking at broadcast filtering **** ENDING LOGGING AT Wed Jan 29 02:59:57 2020