**** BEGIN LOGGING AT Tue Dec 23 02:59:59 2014 Dec 23 03:59:04 Hi all, I have a quick question if anyone can explain some confusing behaviour... $(wildcard */) in a makefile is behaving differently under oe-core vs. manual make invocation and I'm incredibly confused as to why. Dec 23 04:00:38 check bitabke -e yourrecipe to see if B and S are the same. if not, then it's bulding in a separate build directory from the source tree, so that wildcard needs to be fixed to pass the topdir of the source tree Dec 23 04:00:40 at a guess Dec 23 04:01:02 Well, the problem is that $(wildcard */) normally only matches directories. Dec 23 04:01:13 Under bitbake it's also printing out filenames with slashes appended to them. Dec 23 04:02:37 If I manually cd to my makefile directory, run make in a shell with a rule that says echo $(wildcard */), I get a listing of only directories in that folder as expected. Dec 23 04:03:46 If I place the following in a do_install() recipe: cd /same/folder; make; ... It also lists normal files, with slashes at the end. I'm wondering if make is somehow being run in a compatibility mode or with some hidden flags I'm not aware of? Dec 23 04:14:21 Basically... http://pastie.org/9795339 Dec 23 04:32:27 Okay, took strace to it. Seems to be caused by libpseudo. Dec 23 04:49:47 For those interested, pseudo appears to not return ENOTDIR from stat/lstat for "afile/" and instead treats it as if it were passed "afile" :/ Dec 23 04:50:24 Guess I'll go file a bug report on that. Dec 23 04:59:36 ouch, well spotted Dec 23 05:06:39 I don't suppose there's some way I could subdue psuedo for a single command? Dec 23 05:12:33 Oh, PSEUDO_DISABLED=, works for me. Dec 23 07:57:20 I am using raspian and open embedded with toradex boards. The difference is, that I have a reduced busybox on my open embedded system. Is there a easy way to remove the busybox package and use standard debian? Or is this small design the main target of the project without any option to change this? Dec 23 09:19:24 morning all Dec 23 14:30:35 this looks suspicious: ster/src/oe-core/build/tmp-glibc/sysroots/ettus-e300/usr/lib/libstdc++.so -lm -lnsl -pthread Dec 23 14:30:48 why not -lpthread Dec 23 14:31:18 crofton depends Dec 23 14:31:53 I get a link fail on pthread calls Dec 23 14:31:55 weird Dec 23 14:32:02 checking whether pthreads work with -pthread... yes Dec 23 14:38:20 heh, the guy that did the first pass patched out the test directory, now I know why :) **** ENDING LOGGING AT Wed Dec 24 02:59:59 2014