**** BEGIN LOGGING AT Mon Jan 04 02:59:58 2016 Jan 04 10:13:50 all: how can I list the recipes/packages provided by a specific layer? bitbake-layers doesn't seem to offer that.. Jan 04 10:59:30 anyone please: how can I list the recipes/packages provided by a specific layer? bitbake-layers doesn't seem to offer that.. Jan 04 11:00:02 you could grep for .bb file extensions in the layer? Jan 04 11:01:03 or find: find meta-layer -name \*.bb -print Jan 04 11:02:21 bitbake-layers show-recipes ? Jan 04 11:03:24 joshuaglr: that will give you for each layer in your build env Jan 04 11:12:46 bitbake-layers show-recipes list all recipes. I have created an in-house layer that my users will be generally be interested in. so I want to show our own internal targets Jan 04 11:13:40 and yeah find and grep stuff can be used, but not everyone in our team is a linux person and I wanted to have a nice clean (non-hacky) way of doing so Jan 04 11:23:22 Crofton|work: looks like I cannot compile strace-native with 4.11 Jan 04 11:23:32 Crofton|work: but on the other hand, strace 4.11 supports nios2 :) Jan 04 13:52:33 is there a way to specify runtime dependencies for -native packages? Jan 04 13:53:00 RDEPENDS doesn't seem to work and DEPENDS is not strictly true for the recipes I'm writing Jan 04 13:54:20 I get the feeling this use case is not currently dealt with? Jan 04 13:55:53 radhus: RDEPENDS_${PN}_class-native? Jan 04 13:56:10 ah Jan 04 13:56:40 nope :-/ Jan 04 14:35:18 * darknighte nods g'morning Jan 04 16:13:50 all: busybox seems to depend on a package called initscripts. I don't need those rc.d scripts at all. how to get rid of initscripts and update-rc.d dependency in busybox? Jan 04 16:15:27 did you remove sysvinit from your DISTRO_FEATURES? Jan 04 16:31:21 kergoth: I thought I needed sysvinit because I will eventually be using sysvinit like initialization.. Jan 04 16:31:44 if you're using initscripts, then you'll want update-rc.d to set up the rcX.d symlinks, no? Jan 04 16:31:54 if you're using sysvinit, that is Jan 04 16:35:43 kergoth: i see. I think I understand what you mean. I should be using poky-tiny as a template instead of poky in which case I can specify my own VIRTUAL-RUNTIME_init_manager and get rid of sysvinit as a DISTRO_FEATURE. Jan 04 16:35:52 kergoth: am I right? Jan 04 17:49:15 is there a way to escape special characters during a variable assignment? I'm having issues doing a git clone with credentials that have weird chars Jan 04 18:04:05 karobar: its more likely our url handling that's the issue, not variable assignment. what's the failure? Jan 04 18:06:19 ERROR: Fetcher failure: Fetch command failed with exit code 1, output: /bin/sh: -c: line 0: syntax error near unexpected token `)' ... Jan 04 18:19:42 kergoth: the error I just posted happens during this SRC_URI = "git://www.example.com/path/to/repo;protocol=https;branch=master;name=commit;user=sampleName:sample)}{(Password Jan 04 18:20:08 rburton: yt ? Jan 04 18:20:11 did you try passing the user/password in the url, not the parameters? Jan 04 18:20:25 yeah, same issue Jan 04 18:21:08 wonder if it's not correctly quoting the git arguments in the fetcher Jan 04 18:21:13 hmm Jan 04 18:21:22 rburton: have you pulled in the musl phase2 patchset ? since then I have few more recipes fixed. I was planning on sending a phase3 pu Jan 04 18:21:46 that's kinda what i figured Jan 04 18:22:16 karobar: its better to use .netrc to specify username and pw Jan 04 18:22:24 this is true Jan 04 18:22:40 then it's outside the metadata, out of band, and you can lock down its perms more than the recipe Jan 04 18:23:03 okay thanks, I'll give that a shot Jan 04 18:23:06 that way you dont have to clutter the metadata with uname/pw pairs as they might be different for different users I assume Jan 04 18:37:41 yeah, between netrc & an ssh agent, pretty much covers your bases. I wonder if askpass would be correctly spawned if run manually and in the right contexts, though. Jan 04 18:39:02 not sure if its run under fakeroot env but otherwise it should Jan 04 18:39:29 good point Jan 04 19:28:24 i am using smart to upgrade a kernel on a running system Jan 04 19:28:59 the kernel modules are taking a long time Jan 04 19:29:36 lots of packages can take a long time to process.. Jan 04 19:29:41 i currently have 209 kernel modules Jan 04 19:29:47 it's an unfortunate side effect of the resolver being written in pything (in smart) Jan 04 19:31:05 khem: I created the ~/.netrc and added the proper host, user, and pw. running bitbake gave me "fatal: could not read Username for 'https://www.example.com': No such device or address". This error only makes sense to me if bitbake doesn't look at the ~/.netrc file, wouldn't the pseudo user have a separate home directory? Jan 04 19:31:27 i get a message that says Installing, then maybe 15 seconds later, then it says Cleaning kernel-module-, waits 15 seconds and continues Jan 04 19:32:13 fray, are you saying this is a result of the 209 modules, purely? Jan 04 19:32:26 I don't know you'd have to debug/instrument smart.. I do know resolution on kernel-modules, python and perl can be very time consuming due to the number of dependencies.. but the actual install aspect is usually as fast as calling RPM directly Jan 04 19:33:46 the caching and calculation of transactions to take is not too bad Jan 04 19:34:03 the calculation of the transaction is the part I was referring to Jan 04 19:34:22 the 'smart' side of it that is.. what it's going to download and update/remove Jan 04 19:35:17 i thought it might be the depmod calls for the kernel modules, but disabling that as not sped things up Jan 04 19:35:40 look at the kernel modules packages.. are there any scripts (pre/post install/rm)? that could be causing issues.. Jan 04 19:35:49 I've seen that in the past, but I'm not aware of any current issues there Jan 04 19:36:41 kernel-modules-split.bbclass adds a "depmod -a" for each kernel-module package Jan 04 19:36:44 on the target Jan 04 19:37:11 disabling that did not speed things up, unfortunately Jan 04 19:38:11 Ya, that seems very wrong to me Jan 04 19:39:06 do you know of what else smart might be doing in the "Installing" and "Cleaning" steps? Jan 04 19:39:48 smart (at a high level) pulls the repo database down, comes up with a possible solution, then downloads the packages.. generates a transaction and hands off the install to RPM.. Jan 04 19:39:53 thats the high level view.. Jan 04 19:40:15 the possible solution piece and stupid things like packages doing the same thing over and over have been the bottlenecks I've seen.. Jan 04 19:40:48 but you need to instrument it and determine where the problem is happening and if it's inside RPM, if you can replicate it w/o smart (likely you can) which would be a packaging 'issue' of some kind Jan 04 19:41:04 when you run smart there are some arguments to enable verbose and debugging.. I don't remember them off hand though Jan 04 19:41:28 i am using a simple directory of rpm's as a repo, but that would affect the solution calculation, i believe Jan 04 19:41:34 I thought there was also a mode to run though everything but the install itself Jan 04 19:41:55 if you ran in the mode and there was no slowdown, then you know it's something else Jan 04 19:42:07 i have increased rpm-log-level and smart log-level, but i haven't noticed anything yet Jan 04 19:42:38 top doesn't show high cpu usage while smart blocks Jan 04 19:43:30 those are good suggestions: smart without installing, and then just the rpm Jan 04 19:43:39 i'll give those a shot Jan 04 19:44:03 it's been a month or so since I dug into that stuff.. but that is usually how I go about trouble shooting this kind of thing Jan 04 19:44:17 you can similate a smart -> rpm transaction with simply: Jan 04 19:44:25 rpm -Uhv Jan 04 19:44:29 and it usually reproduces the problems Jan 04 19:44:33 * kergoth works on finishing off the damn git shallow support before it drives him nuts Jan 04 20:34:40 b Jan 04 20:35:36 someday I'll learn to not type into xchat during startup while it twitches and lurches between channels. :-/ Jan 04 20:36:02 paulg, it's the beer Jan 04 20:36:20 you havn't had enough.. thus the twitches and lurches.. ;) Jan 04 20:36:27 might be that. :) Jan 04 20:37:29 at least that is a solution that I'm willing to test. Jan 04 20:47:26 khem: not yet, but was planning on that today Jan 04 21:01:56 khem: the fts recipe still has patches with no header/sob/upstream Jan 04 21:39:16 what's the pattern for VARIABLE_append? Is VARIABLE_append_MACHINE valid? Jan 04 21:43:14 it works with overrides. any entry in OVERRIDES can be used in that context Jan 04 21:43:23 see the value of OVERRIDES in meta/conf/bitbake.conf or with bitbake -e Jan 04 21:43:26 but short answer: yes Jan 04 21:44:13 thx Jan 04 21:53:15 tripzero_: the important bit is that _append and _remove is first Jan 04 21:53:46 happy new year all **** ENDING LOGGING AT Tue Jan 05 02:59:58 2016