**** BEGIN LOGGING AT Mon Feb 17 02:59:57 2020 Feb 17 09:53:08 Hauke: ping Feb 17 11:55:34 mangix: please can you use more appropriate branch names? This is all visible in Git and when you are using branch name like "abas" or "abaska" it does not say anything about that. Feb 17 11:57:31 Also is transmission working for you in OpenWrt 18.06? Feb 17 12:45:34 merge via rebase instead of merge would fix that. Feb 17 12:49:18 Yup. It depends if we prefer to do ff-merge or not. Feb 17 13:01:19 who wouldnt prefer linear commit history? Feb 17 13:03:36 github Feb 17 13:03:53 [x] Require linear history Feb 17 13:03:54 done Feb 17 13:04:12 GitLab has this as well Feb 17 13:35:55 merge button - only enable 'allow rebase merging' Feb 17 14:02:16 jow: I've found a bug in the feeds.mk:22; the $(pkg)_*.ipk wildcard will pick up more packages than it should when a package name starts with the same name of a selected package plus an underscore _ Feb 17 14:04:03 This happens to me whenever compiling for mvebu, as kmod-bluetooth is selected by default, and it will pick up kmod-bluetooth_6lowpan. I've tried to come up with a better wildcard, but the version & revision strings are not uniform--having alpha caracters all over. Feb 17 14:05:21 I've looked superficially for cases; there are quite a few of them in the packages feed, and kmod-ieee802154 will also pick up kmod-ieee802154_6lowpan. Feb 17 14:06:20 The easiest solution would be to rename the kmod-*_6lowpan packages using a dash instead of the underscore, and perhaps make this a rule. What do you think? Feb 17 14:15:06 cotequeiroz: yes, wontfix. Packages must not contain underscores in their name as various places (including the one you found) make assumptions about the package format Feb 17 14:15:23 *package name format Feb 17 14:18:22 jow: is there a reason why luci_stats forces rrd stepsize & heartbeat when https://collectd.org/documentation/manpages/collectd.conf.5.shtml#plugin_rrdtool states don't set them unless you have a very good reason to do so? Feb 17 14:18:51 ldir: I simply translated 1:1 translated 12 year old code Feb 17 14:19:03 ldir: might be that things changed, or that it always was wrong to begin with Feb 17 14:19:09 so, likely no particular reason Feb 17 14:19:37 you can change .default = xxx to .placeholder = xxx in places where it is more appropriate to leave the initial value alone Feb 17 14:19:46 Thanks for the input. I'll make the adjustments to the two of them I've found in the main repo right away, and add a note to https://openwrt.org/docs/guide-developer/packages, then adjust the others in the packages feed. Feb 17 14:20:01 cotequeiroz: that would be much appreciated, thank you! Feb 17 14:20:07 hmm, ok Feb 17 14:20:47 ldir: I mena the less config the better, I trust the collectd authors to have a better idea about sane defaults :) Feb 17 14:20:55 jow: what about if you simply don't want a value? Feb 17 14:21:27 ldir: in this case we maybe should consider removing the option entirely Feb 17 14:24:41 ok :-) I'll do some more digging Feb 17 14:25:19 ldir: however I think the "force" aspect came due to the Lua->JS conversion Feb 17 14:25:45 in Lua CBI, .default had a different behaviour Feb 17 14:26:06 am getting some 'real wonky' values in certain graphs and I want to prove I'm not doing something silly first. Feb 17 14:26:14 in JS it will get set as input value when no other value is in the config, which essentially "forces" it as you noted Feb 17 14:26:47 if you change it to placeholder, it will appear as light grey suggestion in the input to give the user an idea of what kind ov values are acceptable but it will not set a vlue if none is given Feb 17 14:27:07 *of, *value Feb 17 14:27:19 placeholder may be good enough.. will try :-) Feb 17 14:28:03 I guess almost all .default options should be converted to .placeholders Feb 17 17:56:25 jow: I've hit another glitch when trying a full build with all packages: I've hit `make[2]: execvp: /usr/bin/env: Argument list too long` in include/package-ipkg.mk:187, and :255, while making `package/luci/{clean,compile}`. Feb 17 18:00:03 A trivial change to `echo ... | xargs` did not work, because make passes the echo statement as arguments to the shell. One solution is to use a temporary file with $$(file >$(TMP_DIR)/$(1).in, $$(call opkg_package_files,$(1)*)) Feb 17 18:03:25 I don't know if that's OK on its own, or if we should try to check its size first with $(words ) to decide whether to run the original command, or use a temp file, to avoid creating a file every time; if so, what threshold should be used? Feb 17 18:06:56 Here's the actual patch I'm using: https://pastebin.com/YeBMKGrX Feb 17 20:45:12 Pepe: i do it for speed reasons. I have a packages directory that I push from and take patches from feeds/packages. Maybe the way forward is to rebase merge PRs... Feb 17 20:50:45 That's wrong behaviour... Feb 17 20:52:28 yeah, I found that almost all .default's aI had were better as .placeholders, so they didnt' save defaults and things too. Feb 17 20:58:30 jow: pong **** ENDING LOGGING AT Tue Feb 18 02:59:57 2020