**** BEGIN LOGGING AT Mon Jun 29 02:59:58 2015 Jun 29 09:48:59 good morning Jun 29 09:51:04 hi mckoan Jun 29 09:56:29 morning all Jun 29 09:58:48 what is the recommended way of supporting HW with addon boards and/or built-in FPGA with OE? At the moment, we create a new MACHINE for each variant, e.g the same board with 2 different FPGA designs. I imagine all these DIY/home-tinkering boards have similar problems, how to support all addon boards/shields/capes etc? Jun 29 10:17:01 mago__: can't you do with one machine that install multiple device trees ? Jun 29 10:33:35 abelloni: sure, i can solve the kernel bits with kernel modules and device trees. but when it comes to image configuration, it gets trickier if you don't have a MACHINE to separate changes. Lets say a addon-board or a FPGA design adds a new ethernet MAC to the system. I can add that to the devicetree, no problem. But how do I for example provide a customized /etc/network/interfaces for a particular variant, Jun 29 10:33:38 without making it a MACHINE? Jun 29 13:15:48 in oe, which package must I install to get ldd on my device... interwebs says glibc but there are many glibc I am confused Jun 29 13:59:12 pompomJuice: believe it or not the package is actually "ldd" :) Jun 29 13:59:33 pompomJuice: but it is provided by glibc -> https://github.com/openembedded/openembedded-core/blob/master/meta/recipes-core/glibc/glibc-package.inc#L41 Jun 29 14:15:03 thanks nrossi, let me check that out Jun 29 17:12:28 anyone goin gto the CCC camp? Jun 29 17:31:54 khem`: is there any recipe which generates an SDK toolchain which generates code to the architecture the SDK is running on? I think it should be something like nativesdk-gcc. Jun 29 17:35:16 mario-goulart: on master iirc yes RPs patch has been accepted Jun 29 17:35:43 khem`: ah, ok. Thanks. Jun 29 17:37:46 mario-goulart: I think its not in Jun 29 17:43:17 khem`: I can't find them. Maybe it is not in, indeed. Jun 29 17:43:48 s/them/it/ Jun 29 20:49:23 oe-classic, my binaries aren't getting stripped. Is this a known/old bug? Jun 29 20:51:14 are they executable? iirc that was a factor, but it's been ages Jun 29 20:51:18 * kergoth shrugs Jun 29 20:52:06 ERROR: runstrip: 'file /home/[...]/sysvinit-2.86-r66.6/package/sbin/init.sysvinit' failed (forced strip) Jun 29 20:52:19 (bad example, that's not an executable) Jun 29 20:55:38 (maybe it is) Jun 29 20:57:15 hmm, failed isn't particularly useful, no messages from the strip command above that? Jun 29 21:01:21 kergoth: OE was trying to use it's own version of "file" to tell if something had been stripped or not already. Jun 29 21:01:21 it's version used my host magic file, which results in a version mismatch and failure, Jun 29 21:01:21 so the binaries aren't getting stripped. Jun 29 21:01:37 easy fix Jun 29 21:29:58 khem`: I changed the following for the UTF-8 support, https://github.com/victronenergy/openembedded-core/commit/b215002513f18b1684a05961bdd4cb2bd3ab97f7 Jun 29 21:34:42 khem`: remains a bit a weird routine, but at least it works... Jun 29 21:35:47 jeroen_: can you submit this patch to busybox ml ? Jun 29 21:36:00 I think its probably a signedness bug Jun 29 21:36:16 what architecture do you use Jun 29 21:36:20 on your target ? Jun 29 21:36:25 is it ppc or mips or arm Jun 29 21:36:28 arm Jun 29 21:37:56 I doubt it is a signedness bug though, I guess the intend is to not print garbage / confuse the console Jun 29 21:38:39 arm then it should not be Jun 29 21:39:00 yes nevertheless submit it Jun 29 21:39:58 mmm, submit a patch to busybox... I don't mind in general, but aren't those guys the folks who sue people? Jun 29 21:43:51 busybox is sometimes used as leverage against folks who ship it on their embedded boards without complying with the gpl, that's about all as far as i know Jun 29 21:54:36 khem`, kergoth: I am not a lawyer, and don't intend to become one, so I prefer to stay away from projects with complicated license policies (and hence busybox) Jun 29 21:56:08 but anyway, the patch fixes the UTF-8 problem (also in master). But might cause some strange characters in the wrong code page instead of question marks.. Jun 29 21:59:55 jeroen_: no, its a poster child Jun 29 22:00:23 due to linksys case Jun 29 22:00:30 its SFLC Jun 29 22:00:42 which was doing it Jun 29 22:01:03 You can bet they would sue an open source project in a heart beat Jun 29 22:01:35 where "they" are eveil closed soure vendors Jun 29 22:01:51 so we respect their licenses, and they respect ours Jun 29 22:06:10 jeroen_: I encourage you to still post it to ml following your corporate opensource policy Jun 29 22:06:17 it will get your patch vetted Jun 29 22:06:30 and may be accepted too if its the right/best fix Jun 29 22:18:11 jeroen_: busybox doesn't have complex licensing.. it's just gpl. the fact that sflc tends to use it as, as khem` says, a poster child, really doesn't say much about the project or its licensing, they could just as well have used any other obiquitous project, it was just easier to use it than the kernel Jun 29 22:18:20 afaik, anyway Jun 29 22:36:21 kergoth, khem: need some rest, will ask, think about it Jun 29 23:02:47 kergoth: I am going to sleep, but there is not "just gpl" Jun 29 23:03:34 kergoth: thre is LGPL 2.? / GPL 2.x? / GPL 3.x Jun 29 23:03:53 kergoth: and likely more Jun 29 23:04:14 kergoth: there is no "the GPL" Jun 29 23:05:02 kergoth: even worse GPL3+ cannot use GPL 2 only Jun 29 23:06:21 kergoth: strictly speaking, it is a mess Jun 29 23:36:52 did you not bother to read the license of the project? busybox is gplv2 Jun 29 23:37:06 not that difficult to get your head around Jun 29 23:37:15 and as khem said earlier, talk to your legal department as appropriate Jun 29 23:38:48 * kergoth wanders off, work day over **** ENDING LOGGING AT Tue Jun 30 02:59:58 2015