**** BEGIN LOGGING AT Tue Oct 27 02:59:59 2015 Oct 27 08:06:19 good morning Oct 27 08:14:29 morning all Oct 27 08:25:52 hi bluelightning Oct 27 08:26:25 hi mckoan Oct 27 08:45:52 mornin Oct 27 12:11:47 hi, when building my image i get "opkg_install_pkg: Package libusb-1.0-0 md5sum mismatch. Either the opkg or the package index are corrupt. Try 'opkg update'.". Even if i rebuild libusb, same problem, any idea ? Oct 27 12:16:56 wakko: sounds like https://bugzilla.yoctoproject.org/show_bug.cgi?id=8578 Oct 27 12:18:26 yes, it happened for other packages as well but a rebuild was enough to fix the issue, there i have no luck Oct 27 12:19:39 by rebuild i mean "bitbake -c cleansstate package; bitbake package; bitbake package-index" Oct 27 12:31:04 wakko: unfortunately if it's the package indexing code that is at fault rebuilding the package won't fix it Oct 27 12:32:06 the issue is that the package indexing code thinks it does not need to update the checksum for the package Oct 27 12:34:28 i get it Oct 27 12:37:41 as a workaround you could move the indexes out of the way and that will force it to regenerate them Oct 27 12:37:55 it would probably be helpful if you could attach a copy of the broken index to that bug Oct 27 12:42:08 where are those indexes located ? Oct 27 12:42:27 sounds like the problem I have every once in a while Oct 27 12:43:02 wakko: tmp/deploy/ipk// Oct 27 12:43:13 ok Oct 27 12:43:23 Jin^eLD: yes, it's been hanging around for a while Oct 27 12:43:40 Jin^eLD: actually we were discussing it yesterday weren't we? Oct 27 12:43:45 could be two recipes outputing the same package name ? Oct 27 12:43:54 bluelightning: yes :) and I entered the issue to bugzilla yesterday Oct 27 12:44:04 Jin^eLD: right, that's your bug I linked to above Oct 27 12:44:07 I had to stop debugging because it started to work Oct 27 12:44:08 wakko: no, we check for that situation explicitly Oct 27 12:44:09 yep, it is Oct 27 12:45:03 wakko: btw mabye you could add a comment to the ticket that you see this problem as well, so it is clear that more people suffer from this issue Oct 27 12:45:12 then i will keep that broken index, rebuild it and diff to see if there is something interesting appart from that hash Oct 27 12:45:53 cool, thanks Oct 27 13:45:27 nerdboy: I was thinking OE built SDK Oct 27 13:47:11 bluelightning: removing the index and rebuilding did not fix the issue... Oct 27 13:48:06 wakko: er... ok - you're sure you removed the right index? (by right index I mean the one for the architecture of the package in question) Oct 27 13:48:18 haaaa Oct 27 13:48:41 no i remove the index of the machine, not index of the arch of the package :( Oct 27 13:49:04 ah ok... take a copy as well, since that's where the problem will lie Oct 27 13:49:10 sorry I should have been clearer earlier Oct 27 13:49:31 no it's logic Oct 27 14:12:49 ok there is strictly nothing noticeable between both Packages files except MD5Sum and Size fields... Oct 27 14:14:13 ha but in the broken version i had a file named Packages.9947 Oct 27 14:14:34 could it be that the indexer script uses glob ? :) Oct 27 14:15:23 this file is incomplete... so it must be a left over from an aborted build Oct 27 18:56:54 bluelightning, ping Oct 27 18:57:07 hi Crofton Oct 27 18:57:27 You've seen what I did in meta-ettus? Oct 27 18:57:37 two layers, each supporting one board Oct 27 18:58:00 this was dumb Oct 27 18:58:20 so moving forward, I want to move all the new boards to the top level Oct 27 18:58:46 so I will end up bsp's and layers in each other Oct 27 18:59:08 will this cause the layer index to explode Oct 27 18:59:11 or toaster/ Oct 27 18:59:15 erm Oct 27 18:59:19 I'm not sure I follow Oct 27 18:59:20 heh Oct 27 18:59:40 I recall you doing the split a while back yes Oct 27 18:59:46 and naming the layers without meta- Oct 27 18:59:48 ;) Oct 27 18:59:50 yes Oct 27 19:00:10 so I wnt to make new boards in meta-ettus Oct 27 19:00:19 but leave the legacy ones alone for now Oct 27 19:00:30 I think it will work form a bitbake pov Oct 27 19:01:11 as long as the top level is also a layer I can't see a problem Oct 27 19:01:16 jsut worried I will make the layer index insane Oct 27 19:01:21 no, it should handle it Oct 27 19:01:24 ok Oct 27 19:01:36 it used to get a bit confused with layers in layers (e.g. meta-intel) but I sorted that out Oct 27 19:01:39 if it explodes, I can look at fixing sooner than I wnat to Oct 27 19:01:41 ok Oct 27 19:01:56 yeah basically, moving to a layer inside layers scenario Oct 27 19:01:56 if the layer index explodes it probably needs fixing :) Oct 27 19:02:13 it won't pick up the structure change by itself though of course Oct 27 19:02:43 I would submit meta-ettus as a new layer Oct 27 19:10:29 Crofton: right, that should work Oct 27 19:22:15 otavio: are the the imx233-olinuxino* machines still "unmaintained"? Oct 27 19:22:33 nerdboy: yes; wish to adopt them? Oct 27 19:23:02 nerdboy: they are in good shape, afaik, as it uses all mainline stuff Oct 27 19:29:13 my boss said she'd buy a couple for testing Oct 27 19:30:47 so yeah, she'll comaintain with me Oct 27 19:34:22 otavio: what's the process? do we need to get freescale tattoo or something? Oct 27 19:38:13 nerdboy: no; just prepare a patch adding your name and become responsive on ml for those ;-) Oct 27 19:39:50 Hi, Oct 27 19:42:59 I currently swith to fido branch and I have a buid issue on mx6q with libgl-mesa-dev required by imx-gpu-viv. I don't a solution, I want build without X. Someone can help me ? Otavio you are fsl maintainer, I think ? Oct 27 19:44:02 the issue: ERROR: Nothing RPROVIDES 'libgl-mesa-dev' (but /home/tperrot/dev/svn/trunk/dist/meta-fsl-arm/recipes-graphics/imx-gpu-viv/imx-gpu-viv_5.0.11.p4.5-hfp.bb RDEPENDS on or otherwise requires it) Oct 27 20:09:12 Sorry, for disturbing you, I found a temporary solution to fix this build issue. good night Oct 27 20:27:00 t0mmy: ?? Oct 27 23:55:28 hmm, how do I force bitbaek to rerun the configure step after I hack the source in work? **** ENDING LOGGING AT Wed Oct 28 02:59:58 2015