**** BEGIN LOGGING AT Tue Mar 28 03:00:03 2017 Mar 28 09:27:26 is there any magic with images that makes them ignore DEPENDS? i've added a native-tool dependency to a bbclass that my image inherits, but it doesn't get built before my image tasks run? Mar 28 09:28:43 i noticed a lot of stuff in image_types.bbclass add tool dependencies to IMAGE_DEPENDS, but that's not in the manual.. Mar 28 11:50:24 mago1: what sort of image are you building ? Mar 28 11:54:38 mago1: what task exactly needs the native tool? Mar 28 11:56:40 jku: Marex: im packaging a rootfs and some other static files into a proprietary archive, the native tool is needed to create the archive. the rootfs is the output of a "normal" OE image, so my recipe depends on that, and the static files are in my SRC_URI. I then just put everything into my "superimage" (e.g the rootfs + static files) and then run the archive tool on that image Mar 28 11:57:15 so it's a kind of image-in-an-image type scenario Mar 28 12:07:48 bird-in-a-bird-in-a-bird Mar 28 12:50:18 RP: :-( ok Mar 28 12:50:28 RP: mesa though is important Mar 28 12:50:57 otavio_: I've merged mesa Mar 28 12:51:14 RP: also eudev please Mar 28 12:51:39 RP: u-boot is not critical Mar 28 12:51:57 otavio: eudev is in too Mar 28 12:52:41 RP: thx! Mar 28 12:54:20 RP: There are few important fixes for etnaviv on the way to mesa 17.0.3; potentially I will backport them Mar 28 12:54:28 RP: once they are merged Mar 28 12:55:19 otavio: right, we can handle that in M4 as things develop Mar 28 15:15:10 mago1: DEPENDS means "our do_configure task depends on the do_populate_sysroot of those tasks", along with the recipe specific sysroot construction. image recipes mark do_configure as noexec, so such depends won't do anything. you'll want to use the 'depends' flag, not the DEPENDS variable Mar 28 15:18:35 kergoth: gotcha, mytask[depends] += "mytool:do_populate_sysroot" seem to work fine Mar 28 15:22:49 kergoth: technically, noexec tasks do have their depends resolved fwiw Mar 28 21:42:39 RP: u-boot is super-important as it fixes the bloody OMAP serial bug ;-) Mar 28 21:47:55 Marex: but its basically too late and would disrupt other layers. I've built M3-rc2 now Mar 28 21:48:46 sounds like time to document a known issue Mar 28 21:48:47 * kergoth yawns Mar 28 21:52:00 RP: even if I say pretty please ? :) Mar 28 21:52:20 RP: well, I kinda knew this answer was coming, so anyway ... Mar 28 21:52:26 kergoth: we can backport a patch too Mar 28 22:06:54 RP: should I send a patch for the missing whoami during kernel build? Mar 28 22:08:05 ant_home: I worry about why that is needed at all :/ Mar 28 22:08:11 ant_home: it really shouldn't be Mar 28 22:08:20 also, why are we not seeing failures due to it? Mar 28 22:08:26 doh..you know you have your name as kernel builder ? :) Mar 28 22:09:00 ant_home: ah. In that case, we should find how we set and override it to something consistent for build reproducibility Mar 28 22:09:42 RP: kernel-source/scripts/mkcompile_h Mar 28 22:14:12 all the others are present (echo, grep, cmp, mv, rm, cut, date, hostname, cat), just whoami missing Mar 28 22:20:28 Marex, the module builds w/out warnings. Now hopeless testings ;) Mar 28 22:20:41 sorry, wrong chan Mar 28 22:25:30 ant_home: I'm quite happy to have my own small fanclub of people who follow me everywhere, hehe Mar 28 22:25:56 Marex: sign me up! :D Mar 28 22:26:20 Marex, I'm always lurking on #oe Mar 28 22:26:42 we even got signs of life from lumag recently Mar 28 22:27:03 bluelightning: are you at OFTC already ? Mar 28 22:27:39 bluelightning: you're a poor fanboy :) Mar 28 22:27:48 Marex: OFTC? Mar 28 22:28:01 bluelightning: debian irc network, there's #mtd and such, hehe Mar 28 22:28:19 Marex: ah right - no, just freenode Mar 28 22:28:22 ant_home: yay :) Mar 28 22:28:30 I'm now poking the last pxa people :) Mar 28 22:29:10 bluelightning: bad fanboi :) Mar 28 22:29:24 bluelightning: things are going well for you otherwise ? Mar 28 22:29:25 Marex: sorry, I'll try to do better :) Mar 28 22:29:33 Marex: yeah, can't complain too much Mar 28 22:29:37 Marex, I did my part on #mtd long ago, when I fixe the NOR driver for collie ... on a defective model :] Mar 28 22:29:40 bluelightning: glad to hear that :) Mar 28 22:30:08 ant_home: I think I still have C1000 and C3100 in my drawer Mar 28 22:30:28 Marex, I'm still a bit ashamed: I bothered for nothing Mar 28 22:31:14 that's why, before sending an unpresentable driver I'd like you or lumag to review...err rewrite it Mar 28 22:31:58 btw I might ask why RP and lunmag Mar 28 22:32:21 di not consider that darn Sharp FTL when writing sharpsl nand Mar 28 22:32:42 (if they remember 10 years later) Mar 28 22:33:49 ok too late here.. good night to all Mar 28 22:33:51 "why RP" sounds like a good question to ask :) Mar 28 22:33:58 he he Mar 28 22:34:21 I don't think it was because of patents, anyway Mar 28 22:34:41 no, there was some good reason. I don't think we needed/wanted an FTL Mar 28 22:35:14 I too don't think we need the full FTL in kernel Mar 28 22:35:28 just th few bits for reading the mtdparts from nand Mar 28 22:35:49 you only need the ftl if you want to do fat on the block device. We were happy with jffs2 Mar 28 22:36:53 ant_home: I am NOT rewriting anything, please submit patches Mar 28 22:37:13 now joking, Im opying the style from the other drivers Mar 28 22:37:42 but I'm really a bit confused, the partitioning is done by platform AND board Mar 28 22:38:17 I have to study more...tomorrow Mar 28 22:38:22 good night **** ENDING LOGGING AT Wed Mar 29 03:00:03 2017