**** BEGIN LOGGING AT Thu Nov 28 02:59:58 2019 Nov 28 05:17:23 What is the reg property in the ath9k devicetree binding? The documentation just says "Address and length of the register set for the device". How do I determine the correct value for my device? Nov 28 10:07:15 hi all anybody here who maintains cmake based package for openwrt? Nov 28 10:08:57 i'm trying to package avro-cpp, which needs for tests avrogencpp which generates cpp/hpp files from schema Nov 28 10:09:42 which in simple makefile is built for target platform Nov 28 10:10:15 not for host platform Nov 28 10:12:09 is there any cmake based package makefile which has also host install part? Nov 28 10:32:14 yes my softether5 package does this Nov 28 10:32:26 check how i build hamcorebuilder on host Nov 28 10:32:40 andy2244, +1 thanks :) will check Nov 28 10:33:04 andy2244, in packages repo? Nov 28 10:33:14 yes in master packages Nov 28 10:38:58 andy2244, "include $(INCLUDE_DIR)/nls.mk" line is for language support? Nov 28 10:39:09 yes Nov 28 10:39:34 i think the iconv stuff macros Nov 28 10:39:51 i see Nov 28 10:42:35 btw cmake/make works similar for host utils, you define all the host variables/stages and just build the tool you need for host, this way you can define host only depends Nov 28 10:42:37 andy2244, thanks a lot this is exactly what i need :) Nov 28 10:43:08 the trick i use is just build the host util statically, so i dont have to worry about shared libs Nov 28 10:44:15 this works this way: Nov 28 10:44:16 HOST_BUILD_DEPENDS:=ncurses/host readline/host Nov 28 10:44:16 PKG_BUILD_DEPENDS:=softethervpn5/host Nov 28 10:44:27 andy2244, by default avrogencpp is using static libraries Nov 28 10:44:29 i mean for the depends Nov 28 10:44:48 than you are gtg Nov 28 10:45:14 yes Nov 28 10:45:26 once more - thanks Nov 28 10:45:31 np Nov 28 10:50:13 doh, found _one_ of my buildroots wasn't set to the same dl directory, so removing files to force a new download wsa't _reallllly- working there. Nov 28 10:58:25 :) Nov 28 10:58:38 karlp: do you use the cleanup script? Nov 28 10:59:00 what cleanup script? Nov 28 10:59:05 scripts/dl_cleanup.py Nov 28 10:59:39 ah, no, that's not the issue, Nov 28 10:59:54 this is where the version is "HEAD" so it doesn't download it again Nov 28 10:59:59 ok Nov 28 11:00:48 for dl_cleanup sort of things I just every now and again just delete the whole tree :) Nov 28 11:01:03 :P Nov 28 11:01:32 somewhere in my 19.07 builds ipv6 on LAN seems to have been broken :-/ Nov 28 11:01:33 great Nov 28 12:47:26 blogic: https://bugs.openwrt.org/index.php?do=details&task_id=2615 (wifi reconfig issue) Nov 28 12:48:19 maybe I should report my issue caused by reconf there as well Nov 28 12:48:23 client devices are no longer reliable Nov 28 12:49:03 if for some reason association doesn't work immediately, the interface isn't brought up by netifd and dhcp doesn't work Nov 28 12:50:00 which apparently is also a problem if the client boots faster than the AP (e.g. after a power outage) Nov 28 12:55:01 Borromini: you've forget to add which device you're testing ramips 4.19 on :) Nov 28 13:07:19 jow: seems like we've some buildbot issue Nov 28 13:08:09 probably only on master Nov 28 13:18:06 does ramips 4.19 include the upstream pcie driver that Rosen backported some time ago? Nov 28 13:18:54 ynezz: yes - sorry Nov 28 13:19:13 was updating the post just after you commented on the thread ^_^ Nov 28 13:44:09 ynezz: sysupgrade seems broken on my rt-ac57u though, strangely enough not on the dir-860l Nov 28 13:45:26 someone on github commented on https://git.openwrt.org/5f683339 to ask to backport this change to 19.07 and 18.06. do we want to introduce new symbols in release branches? Nov 28 13:46:20 19.07 is not out yet, so it should be ok Nov 28 13:46:30 good fix Nov 28 13:46:49 for 18.06 I'm not sure, probably don't care? Nov 28 13:47:02 I do not :) Nov 28 13:47:16 ok I'll cherry-pick it to 19.07 Nov 28 13:47:28 its almost EOL, we cant rule out any potential issues, this is probably not so used config option Nov 28 13:47:41 so not worth the hassle/risks Nov 28 14:54:14 who decided it was a great idea that PKGARCH should go go as a field in the "define Package/blah" with NO _, (and a PKG prefix at all) but alllll the other PKG_BLAH have an underscore :| Nov 28 15:18:20 git has the answer :) Nov 28 15:25:55 yeah, well it can't be changed now without huge swathes of compat either. Nov 28 15:26:00 just... bleh Nov 28 15:27:19 well, you could have both variables for a transition period, but it wouldn't be fun Nov 28 15:29:18 goes back to 2010 anyway, so it's more that it wasn't used much in the past Nov 28 15:29:44 kristrev: i don't think so Nov 28 15:30:00 mt7621 pcie upstream driver seems to be buggy Nov 28 15:31:01 AFAIK even on the latest kernel it has some hard to debug issue Nov 28 15:40:07 hi can anybody take a look on https://github.com/openwrt/openwrt/pull/2576 ? Nov 28 15:41:57 what is proper way to get such change to 18.06 and 19.07 branches? Nov 28 15:42:41 indy: explaining why it should be backported in the PR Nov 28 16:02:37 adrianschmutzler, so abandon 2576 and make another one with prefix? Nov 28 16:04:04 adrianschmutzler, so cherry-pick that to both branches? Nov 28 16:15:29 why doesn't ar71xx tiny target try to install kernel metapackage? https://pastebin.com/NhpnpcM1 Nov 28 16:21:47 adrianschmutzler, 2577 2578 and 2579 :) Nov 28 16:28:46 oh no, now we have three clones Nov 28 16:31:54 indy: I implied that you should _argue_ why it should be backported, not just create the same pull request three times Nov 28 16:31:55 mangix: Great, thanks for letting me know. I have not had time to test yet, just remember all the mess it caused when the backport was merged Nov 28 16:35:58 adrianschmutzler, thanks for clearance, commented Nov 28 19:46:33 kristrev: I believe that backport required a bunch of DTS changes. The hardware that I used did not device any devices on the PCIe bus (they were all SATA controllers, no wifi). Nov 28 19:46:43 *did not have Nov 28 21:13:24 fyi, I reported the breakage caused by the hostapd ubus reload changes: https://bugs.openwrt.org/index.php?do=details&task_id=2637 Nov 28 21:46:42 stintel: ynezz: It was me who asked https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=5f6833395293548f9fdf4897d9766417f2990bac to backport it to openwrt-18.06 branch. Sorry, but the symbol is just there https://github.com/openwrt/openwrt/blob/openwrt-18.06/config/Config-kernel.in#L737 but it has the same typo, which was discovered recently. So you are leaving it broken even you Nov 28 21:46:44 know about it? I am not sure that it is a good idea! Nov 28 21:51:59 Pepe: it's a new symbol due to the rename, in 19.07 it's fine as we're pre-release but it doesn't seem like a good idea for 18.06 Nov 28 21:56:19 it's purely cosmetic so it doesn't break anything, changing it *might* result in people having the old one enabled not having the new one enabled anymore Nov 28 22:04:46 Many other things for 18.06 are not good idea (check the latest cherry-picks), which leads in reboot loop and saying that it is almost EoL, it is not good idea as well as it is the only stable branch which OpenWrt have. Nov 28 22:07:58 mangix: nitpicking now but if you separate those parameters to each on their own line, order them alphabetically while at it ;) Nov 28 22:19:41 i just received an invitation to gitlab.com/openwrt and i'm a bit surprised we are dealing with them. not that i like github, but at least they don't have a legacy of deliberately ruining other FLOSS projects (-> gitorious.org) at the cost of the users of that other service (gitorious.org went offline without prior notice) Nov 28 22:23:44 dangole: for now it's for evaluation purposes only and CI Nov 28 22:24:35 dangole: we've started the evaluation in the underground, on the private instance of lynxis, but people complained that we're not doing it in public Nov 28 22:25:35 dangole: so we moved in the open to gitlab.com, because we can't have testing self-hosted gitlab instance under openwrt.org domain for the obvious reasons Nov 28 22:28:09 and I remember your words about gitorious from Hamburg meeting very well Nov 28 22:31:41 https://gitlab.com/openwrt/gitlab-evaluation/issues/5 (integrate mail based workflow) is probably the last issue we(myself and Paul) would like to get solved in order to start discussion on the mailing list and possibly vote Nov 28 22:33:47 when I talk about gitlab, it's always about self-hosted one, not that one on gitlab.com currently, this can probably stay as it is, as some kind of mirror/backup, or removed, whatever Nov 28 22:35:11 and I've just started adding CI to ucert, and noticed from commit, that you've account on GitLab and thus added you to OpenWrt group Nov 28 22:41:47 ynezz: thanks for the clarification, i'll sleep much better now ;) Nov 28 22:43:00 lol, good night then Nov 28 22:51:55 hmmm, why would reinstall a package overwrite the config file in /etc/config/ when that config file is added to that package its `define Package/foo/conffiles` ? Nov 29 00:13:20 stintel: bad backport Nov 29 00:13:42 mangix: ? Nov 29 00:13:54 unless i'm thinking of something else Nov 29 00:15:43 * mangix wonders who uses HFS with OpenWrt Nov 29 00:16:19 haha, HFS+ even, which afaik does not even support read-write Nov 29 00:17:14 I think HFS+ is on life support currently Nov 29 00:17:40 macOS uses APFS. No driver for Linux AFAIK Nov 29 00:25:51 I will note that hfsprogs in the packages tree is outdated. **** ENDING LOGGING AT Fri Nov 29 02:59:58 2019