**** BEGIN LOGGING AT Fri Nov 15 02:59:58 2019 Nov 15 06:22:45 build #48 of ramips/rt3883 is complete: Failure [failed images] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/ramips%2Frt3883/builds/48 blamelist: Leon M. George , Sungbo Eo , Jo-Philipp Wich , Eneas U de Queiroz , Kyle Copperfield Nov 15 06:22:45 , Hauke Mehrtens Nov 15 06:35:28 let me guess, hostapd 2.9 too big? Nov 15 06:42:34 i just built libpcre with jit enabled and it works just fine on arm a15 Nov 15 06:43:19 benchmarks suggest it should be faster by a factor 10 so maybe we should deploy --enable-jit by default on a known-good set of archs Nov 15 06:44:57 i will conduct more tests and create a pr in the coming days if things go according to plan Nov 15 06:46:55 oh and i will stick with pcre(1) since it is a dependency of wget and that's installed by-default basically everywhere Nov 15 06:53:48 good, just simple hiccup during backporting Nov 15 08:43:11 morning Nov 15 08:50:49 blogic: correct! - you have passed the test - you're cleared to code Nov 15 08:51:33 unlike me who can't even get his nick right Nov 15 08:52:03 that's better Nov 15 08:54:07 In a logic bomb style and to throw another voice into the opinion pile - I cannot think of a reason why an "announce-list" is not a good idea Nov 15 09:51:22 ldir: me neither. Having no bugs and fast development is a good idea as well, it just has to be done by someone Nov 15 09:52:07 jow: point taken Nov 15 09:54:08 jow: but I absolutely agree having the list and then not using it worse than not having the list. Nov 15 09:55:05 ldir: I can only speak from experience with things like release announcment or release notes Nov 15 09:55:20 experience has shown that *nobody* steps up to do it Nov 15 09:55:28 its all just *shrug* back to coding Nov 15 09:55:28 gladiac1337: actually, wget _isn't_ installed by default everywhere :) Nov 15 09:55:58 jow: it's standard with 'documentation' Nov 15 09:58:11 Just checked out 19.07 branch and having problems to build it: Nov 15 09:58:12 scripts/kconfig/conf --silentoldconfig Kconfig Nov 15 09:58:12 net/sched/Kconfig:44: warning: menuconfig statement without prompt Nov 15 09:58:47 and guestions for kernel config begin Nov 15 10:03:07 make menuconfig, diffconfig etc. also warn about recursive dependencies, what is going on with this? Nov 15 10:07:41 if you were in another branch before, you sometimes have some cleaning to do. Nov 15 10:21:06 build #36 of ramips/rt3883 is complete: Failure [failed images] Build details are at http://buildbot.openwrt.org/openwrt-18.06/images/builders/ramips%2Frt3883/builds/36 blamelist: Adrian Schmutzler , Sungbo Eo Nov 15 10:21:09 karlp: i guess you are probably right. 4mb and 8mb devices probably wont have it. also - i just learned that pcre2 supports jit for more architectures than the old pcre(1). furthermore it has the newer api and according to the developers should be used whenever possible. Nov 15 10:22:08 busybox wget won't have pcre, and we normally have what is it again, uclient-fetch? symlinked as wget anyway. Nov 15 10:22:53 i see Nov 15 10:22:59 I don't know what you're working on with pcre really, just saying that "wget uses it" isn't really relevant :) Nov 15 10:23:53 is there an elegant way to ifeq several targets in Makefiles? using ifeq ($(CONFIG_TARGET_BLAH),y) is kinda messy if you have 20 targets you want to whitelist Nov 15 10:44:52 blogic: You seem to have cut a path in your mt7629 commit Nov 15 10:45:07 https://github.com/openwrt/openwrt/commit/50735df7b238ce886c6f718b3e9437a09c8efb28 Nov 15 11:05:16 jow: so I might be otu of luck here, but with the class cbi-tabcontainer, that automatically makes a nice
    for the tabs to select them, I can't get that to play nicely with dynamic tabs based on a knockout model. I need to somehow tell things to "update" again once the knockout model is all loaded. Nov 15 11:06:41 I had been manually creating tab menus, and using the old bootstrap classes and bootstrap.js to get the clicking on them working, but it's all b0rked in 1907 doing it that way. I've got rid of most of it, but dynamic set of tabs is still something I can't figure out Nov 15 11:12:32 karlp: https://github.com/openwrt/luci/blob/86f492173d1daab8b75e4d4bd70ba06872ea70fa/modules/luci-mod-system/htdocs/luci-static/resources/view/system/startup.js#L125 Nov 15 11:13:26 make a container element with a bunch of child elements having "data-tab": "foo", "data-tab-title": "Whatever", then call ui.tabs.initTabGroup() on the containers child elements Nov 15 11:15:39 yeah, I've got my data-tab and data-tab-title, ok, I'll try calling init manually when it's done. Nov 15 11:16:33 auto-init only works for server side rendered html, which is complete before luci.js even loads Nov 15 11:19:21 any tips on how to make the 'require ui' stuff work? Nov 15 11:34:04 karlp: depends on the context your code is executed in Nov 15 11:34:13 karlp: is it loades as view() ? Nov 15 11:34:16 *loaded Nov 15 11:35:17 template(blah) and then the .htm file script tags to my own js. Nov 15 11:35:27 ah Nov 15 11:35:29 hm Nov 15 11:36:02 I've got a bunch of things like this https://zerobin.net/?76d7f5c4564b05e6#3J/sVEhevC4nSgXFI4KkkoteysGcME+SYgGyLoB7ukA= for gettingl uci template access that I can then get to from js, Nov 15 11:36:06 could I get to ui there? Nov 15 11:36:07 then use L.require('ui').then(function(ui) { /* your code using ui */ }) Nov 15 11:49:29 ok. its trying to do the right thing now, I get a second UL generated, but it contains a single broken element. am I poitning at the wrong container? should I be passing some sort of selector of _just_ the dynamic tabs? https://zerobin.net/?c5f11185529adb34#egpvEjgfZBPqGms9kwfKShhaCvjrIC0rlirXtANDzN4= Nov 15 11:58:22 ok, with a selector on the group I can get one tab section for the dynamic and one for the static. Nov 15 11:58:26 so getting somewhere :) Nov 15 12:05:30 ok, I've got it working foradding my static tab to a dynamic list and registgering it and geting a nice single tablist. Nov 15 12:06:18 however, I still always have the static tab at the bottom with it's own single element tablist, it seems it's done it automatically just on seeint the data-tab and data-tab-title tags, even though I removed the cbi-tabcontainer from the outer div. Nov 15 12:06:27 is there anyway of preventing the static automatic tab generation? Nov 15 12:19:43 puttting data-initialized=true on it means it gets the title in the dynamic list, but no tab content. Nov 15 12:36:13 karlp: could you share yome code and screenshots? Nov 15 12:38:02 karlp: so from what I gathered the automatic initialization races with some delayed code of yours, but I am having trouble understanding the details Nov 15 12:38:19 yeah, will package something up. one tick Nov 15 12:44:56 https://zerobin.net/?e078deca80649de9#mEM21RhZ/QvG5fD7g43laXoJq/v3tJN0J2Rsg4JCRW0= Nov 15 12:45:07 images are in https://imgur.com/a/w4vX3cB Nov 15 12:45:27 I gotta get some lunch before the kitchen closes though, back in a bit Nov 15 12:48:22 karlp: you need to move the `data-tab="newconnection"` div into the same parent div as the ko foreach macro Nov 15 12:48:54 karlp: and you need to somehow hide the "data-tab" so that it is not seen at load time Nov 15 12:50:41 karlp: best would be to add `data-tab="newconnection"` using setAttribute() right before ui.tabs.initTabGroup() Nov 15 13:15:20 what a headline: https://news.ycombinator.com/item?id=21543038 Nov 15 13:15:43 It's not even merged yet. Nov 15 13:26:15 yeah, I had them in the same div originally, that seemed best, but that just gives you a single tab of "new connection" Nov 15 13:26:29 I'll try manually adding the tag before updating, Nov 15 13:28:31 taht works, feels kinda yuck, but yeah, I should have tried that earlier, adding it manually afterwards to avoid the setup. Nov 15 13:28:36 thanks a lot Nov 15 13:35:47 mangix: thats how marketing works Nov 15 13:43:13 The guy asked for a backport to 19.07. I guess I should have said no instead of maybe. Nov 15 13:44:33 considering that it would only be in feeds anyway, the distinction is minor Nov 15 13:45:07 jow: eh, adding and removing items from my dynamic list doesn't work now. I've tried calling ui.tabs.updateTabs($(".tabdynamic")); where I would handle the add/remove, but doesn't help much. Nov 15 13:46:48 you want to dynamically add/remove tabs? Nov 15 13:48:19 yes :) Nov 15 13:48:42 I could jsut go back to the bootstrap js things I had before, it was really just styling that was all horribly broken. Nov 15 13:48:50 I could probably fix that in other ways. Nov 15 13:52:03 the "new connection" tab is meant to add a tab into the set, and each tab has a "remove connection" button that drops it from the list Nov 15 14:01:21 blogic: hey, it seems you just added diff file named target/linux/mediatek/patches-4.1 Nov 15 14:01:32 blogic: patches-4.1 is a name of file from what I can see (not a directory) Nov 15 14:08:22 rmilecki: To me it looks like its just been cut off there Nov 15 14:08:43 So, originally it might have been some target/linux/mediatek/patches-4.19/something, too Nov 15 14:08:50 oh, so maybe it's git-scm issue Nov 15 14:09:10 i was looking at https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=50735df7b238ce886c6f718b3e9437a09c8efb28 Nov 15 14:09:43 no... it still looks wrong there: https://git.openwrt.org/?p=openwrt/openwrt.git;a=tree;f=target/linux/mediatek;h=868c85ec1bc3a9de06ec6d183c5b0293397efc6f;hb=HEAD Nov 15 14:09:48 No, you are right, I was just speculating for the reason Nov 15 14:12:11 blogic might be busy with something else, I also posted a patch comment in GitHub some hours ago Nov 15 14:19:34 jow: does my luci language tweak look ok? Have I done the right thing at least in terms of tweaking the same source string in all the PO files ? Nov 15 14:20:31 ldir: yep, it looks fine Nov 15 14:21:07 ok, I think I'll squash those all into 1 commit Nov 15 14:21:31 and update the commit message :-) Nov 15 14:29:28 ldir: maybe hint at the fact that you're a native speaker / that it is the proper form for native english Nov 15 14:35:01 hi Nov 15 14:35:10 i did drop that file in the commit after did i not ? Nov 15 14:35:37 blogic: there is no commit after?! Nov 15 14:35:48 oh Nov 15 14:35:52 i forgot to push as usual :-) Nov 15 14:36:14 sorry about that Nov 15 14:36:25 it was a stray file from when i was diffing diffs a month ago :-) Nov 15 14:37:26 well, other people push stray characters to stable branches, so don't worry ;-) Nov 15 14:37:53 jow: better? Being a native is no guarantee that I really know how to speak the language anyway :-) I just felt the original had a somewhat Germanic base to it Nov 15 14:41:18 The 'Can I get a cup of coffee?' thing is an American import, prompting the grumpy in me to ask in return "I don't know? Can you?" - You can possibly have a cup of coffee, or maybe even "may I have a cup of coffee" - but you'll find some who regard "Can I get a..." as a borderline stabbing offence :-) Nov 15 14:46:16 ldir: let me find that pr Nov 15 14:47:14 ldir: much better - merged. Nov 15 14:51:43 Great - thank you - that's been niggling me for months :-) Nov 15 14:51:54 what is the mechanism to adjust the kernel configuration options depending on the "features" that are enabled ? Nov 15 15:59:11 build #49 of ramips/rt3883 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/ramips%2Frt3883/builds/49 Nov 15 16:47:39 jow: ok, different things... :) using s:taboption() and a add/removable TypedSection, only fields from the first tab of each instance of the section are saved? Nov 15 16:48:53 karlp: that sounds odd. Is this client side (form.js) or server side (cbi.lua) ? Nov 15 16:49:34 client side, this is some brand new stuff Nov 15 16:49:41 https://zerobin.net/?69d6fde86753189b#eJl+2BJ7j4FUqRNw2UhxqAeXuT98L8Ys0TlzZNbLH5Y= Nov 15 16:52:07 bah. must be illegal option names with - in them. Nov 15 16:52:30 it was just saying, "no chnages to apply" and clearing the form everytime I tried to save Nov 15 16:52:54 what is the o.modalonly stuff for? Nov 15 16:54:16 it is used by TableSections and GridSections Nov 15 16:54:41 these will only display a certain amount of options per row by default and offer an "Edit..." button for each row whcih opens a modal with the remaining options Nov 15 16:55:02 .modalonly = true means the option is only rendered in the modal popup Nov 15 16:55:13 .modalonly = false menas the option is only rendered in the row Nov 15 16:55:20 undefined means its rendered in both places Nov 15 16:55:53 see e.g. network -> static routes for a live example Nov 15 16:56:07 for TypedSections it has no effect and can be omitted Nov 15 16:58:08 also network -> interfaces and network -> wireless are actually GridSection Nov 15 17:01:09 that's where that o.cast = table goes too I guess. Nov 15 17:02:09 o.cast = table is not recognized anymore. If you need to force the input to list or option notation, you should overwrite the .write function Nov 15 17:02:36 usually it is inferred from the underlying widget Nov 15 17:02:50 e.g. DynamicList, it will store as uci list Nov 15 17:03:03 just saw tthat when I was looking at other stuff, trying _not_ todo special things :) Nov 15 17:04:12 about the invalid option names with dashes, that's a good catch actually. Iirc form.js does not validate that at all, it will pass it on to the underlying uci.js which will silently filter it away Nov 15 17:04:18 hence the "No changes" message Nov 15 17:04:56 sounds like hwat I was seeing Nov 15 17:08:26 Time to leave the office. bbl Nov 15 17:38:53 jow: yeah, me too :) thanks for all your help today! Nov 15 18:35:55 Hauke: does Lantiq ASE only support mips16? Nov 15 19:02:31 How do I build a package off Github that requires running bootstrap before configure? is there a PKG_FIXUP to add to the Makefile? - https://github.com/duosecurity/duo_unix Makefile - https://gist.github.com/Strykar/fc7820924c50894801153bb29a4bd8a6 Nov 15 19:04:21 Strykar: you can overwite the Configure section in the package Makefile and call the bootstrap there and then call the Configure/Default Nov 15 19:04:50 mangix: let me check Nov 15 19:08:00 Actually I don't think mips16 is it. mips_mips32 has issues compiling code that uses the pause ASM instruction. That was introduced in mips32r2 I believe. Nov 15 19:10:20 mangix: do you get compile problems? Nov 15 19:11:39 Hauke: https://downloads.openwrt.org/snapshots/faillogs/mips_mips32/packages/boost/compile.txt Nov 15 19:11:58 Hauke, thanks, trying that now Nov 15 19:13:12 Error: opcode not supported on this processor: mips32 (mips32) `pause'Error: opcode not supported on this processor: mips32 (mips32) `pause' Nov 15 19:13:14 mangix: the Amazon SE (ASE) support MIPS32R2 and mips16e Nov 15 19:13:25 it is a 4KEc CPU Nov 15 19:13:37 Ah so maybe this is for a different target Nov 15 19:14:29 I just grepped CPU_TYPE in target/linux Nov 15 19:14:45 but the target defines CPU_TYPE:=mips32 Nov 15 19:14:53 so it does not for mips32r2 Nov 15 19:15:01 but for mips32r1 Nov 15 19:15:57 Strykar: don't use that bootstrap file, use PKG_FIXUP:=autoreconf instead Nov 15 19:16:07 lantiq/ase is the only one that defines mips32. All the others are either 24kc oor 74kc Nov 15 19:16:21 mangix: ok then we should probably also change this Nov 15 19:16:36 but I there is also a problem in boost Nov 15 19:16:53 yes there is. I submitted a patch upstream Nov 15 19:17:01 ok thanks Nov 15 19:17:14 which I probably need to fix. But I can't reproduce the build failure with GCC9 Nov 15 19:17:21 trying with GCC8 now Nov 15 19:17:35 Hauke: does that mean that we can get rid of the mips32 pkgarch? Nov 15 19:17:54 by switching lantiq/ase to 24kc, or did I misunderstood? Nov 15 19:17:59 if the doc I have is correct yes Nov 15 19:18:09 I do not have the hardware at hand Nov 15 19:20:00 jow, is there an existing package that uses bootstrap I could look at? Nov 15 19:20:23 Strykar: you don't need that bootstrap file, it just calls autoreconf in the end Nov 15 19:20:31 PKG_FIXUP:=autoreconf will do the same, but properly Nov 15 19:21:12 if you call the bootstrap file it will try to invoke wrong versions of autotools Nov 15 19:22:03 oh no, wikidevi is down Nov 15 19:22:26 archive.org has it i believe Nov 15 19:23:19 https://archive.org/download/wikidevi Nov 15 19:24:16 Strykar: for that particular package (duo_unix) you'll need PKG_FIXUP:=autoreconf and PKG_MACRO_PATHS:=autotools Nov 15 19:24:31 Strykar: to mimick "exec autoreconf -vfi -I autotools" performed by the bootstrap script Nov 15 19:24:33 http://en.techinfodepot.shoutwiki.com/wiki/Main_Page Nov 15 19:24:48 https://web.archive.org/web/20191022065144/https://wikidevi.com/wiki/Main_Page Nov 15 19:26:59 wait a minute...that buildbot error is bogus Nov 15 19:27:18 boost-fiber should be excluded. Nov 15 19:30:59 no wonder I couldn't replicate Nov 15 19:58:34 Trying to build openwrt without the ath10k-ct drivers. Im removing the drivers from make menuconfig but they keep coming back. Any idea? Nov 15 19:59:05 "keep coming back" ? Nov 15 19:59:06 Trying to replace with the ath10k (non-ct) Nov 15 19:59:20 karlp: After I run make the complation puts them back Nov 15 20:00:26 And thus the compilation fails because ath10 and ath10-ct cant be installed at the same time Nov 15 20:05:04 what target/branch? Nov 15 20:06:04 IPQ40XX, GLINET B1300 Nov 15 20:06:29 seems to be because I run make defconfig after I paste in my diffconfig Nov 15 20:06:54 I was taught to always run make defconfig after copying over your customer .config, but I guess Im wrong Nov 15 20:07:24 customer==custom* Nov 15 20:07:33 depends what's in your config snippet Nov 15 20:07:38 you normally do, if it's a snippet Nov 15 20:07:45 if it's a full config you migth get some different things Nov 15 20:07:51 but you run menuconfig _after_ defconfig Nov 15 20:08:04 and turn it off there, and you should be fine. Nov 15 20:08:18 I've not looked at it's config options in any detail to tell you what to put in your snippet Nov 15 20:08:52 My snippet is scripts/diffconfig.sh | grep -v "#" > Nov 15 20:09:08 where did you learn to do that? Nov 15 20:09:34 # BVLAH IS NOT SET _isn't a comment_ Nov 15 20:09:38 yyou can't just strip them all out Nov 15 20:10:18 I did that to have a cleaner git and then I run make defconfig, the # comes back Nov 15 20:11:41 well, do menuconfig _last_ and turn it off, and you get the behaviour you want right? Nov 15 20:12:00 from there you can figure out on your own how to make your own config snippets to be expanded with defconfig Nov 15 20:12:14 yeh I guess, but I need to be scriptable so Ill have to remove what I want to remove with sed instead Nov 15 20:12:35 I just think you're on the wrong track trying to remove things already Nov 15 20:15:27 barhom: karlp is right, "# CONFIG_bla is not set" is meaningful and means "the config symbol bla is set and has the value 'n'" - absence of this line means "the config symbol bla isn't set", which means it will get its default value when running defconfig/menuconfig Nov 15 20:17:17 KanjiMonster: that makes a lot of sense. Ill start saving the # CONFIG_bla Nov 15 20:35:30 karlp, KanjiMonster, thanks compiled successfully Nov 15 20:52:39 basically, just don't do that hoel grep/sed game. Nov 15 20:52:53 just /scripts/diffconfig > mystub.config **** ENDING LOGGING AT Sat Nov 16 02:59:57 2019