**** BEGIN LOGGING AT Sun Dec 18 02:59:57 2011
Dec 18 18:34:52 <Jay7> ant____: hi
Dec 18 18:35:17 <Jay7> ant____: what do you think about increasing default cmdline buffer size to 1024
Dec 18 18:35:36 <Jay7> in worst case our cmdline would be truncated by kexec'ed kernel
Dec 18 18:37:19 <Jay7> second thing I'm considering - renaming boot.cfg to something other..
Dec 18 18:37:31 <Jay7> just to not interfere with PMON
Dec 18 18:40:22 <Jay7> oh.. here are even more projects with boot.cfg file
Dec 18 18:40:30 <Jay7> vmware mboot
Dec 18 23:13:47 <ant____> Jay7: sorry, I'm back now
Dec 18 23:15:02 <ant____> about buffer size, no problems for some bytes more
Dec 18 23:16:13 <ant____> about renaming it, well, I don't see a strict necessity
Dec 18 23:16:46 <ant____> btw I'll have to push some fixes for boot.cfg for meta-handheld
Dec 18 23:17:21 <ant____> bluelightning: Paul, are you around?
Dec 18 23:17:30 <bluelightning> ant____: yep
Dec 18 23:17:36 <ant____> hi
Dec 18 23:18:38 <ant____> as above, there is a $DISTRO var in boot.cfg for easy naming of the label. OFC if yoy build distroless the var remains unexpanded :/
Dec 18 23:19:26 <ant____> then I have to solve the infamous issue of /boot/zImage symlink missing
Dec 18 23:20:03 <ant____> I was thinking to add it to kexecboot_cfg but we don't always Rdepend on kernel, do we?
Dec 18 23:22:01 <bluelightning> ant____: hmm, I'm not sure what the mechanics of that last one should be
Dec 18 23:22:35 <ant____> exactly...
Dec 18 23:22:41 <bluelightning> ant____: as for $DISTRO, that's a bit of a bummer... you might have to get fancy and use some python logic to set a different value if DISTRO is unset
Dec 18 23:24:05 <ant____> about symlink, if we create it on image_rootfs_postrprocess stage we solve only part of the issue: on kernel upgrade the symlink will be stale
Dec 18 23:24:19 <ant____> ...but we deliver bootable images
Dec 18 23:24:38 <bluelightning> seems like kernel postinst should check for the existence of that symlink and update it if present...
Dec 18 23:24:50 <bluelightning> how come we haven't handled that before?
Dec 18 23:25:04 <bluelightning> we've always had a symlink there AFAIK
Dec 18 23:25:08 <ant____> sure
Dec 18 23:25:22 <ant____> I'm wondering 'cause there are no complaints
Dec 18 23:25:42 <ant____> and a rapid check revealed that e.g. Angstrom and Yocto do not add the symlink
Dec 18 23:25:54 <ant____> (very random grep)
Dec 18 23:26:06 <bluelightning> historically we've been very poor at handling upgrades via the packaging system, so a lot of people don't use that path I guess
Dec 18 23:26:51 <ant____> the oinly trace is the 'broken' code in make_zimage_symlink_relative() in image.bbclass
Dec 18 23:27:19 <ant____> but grep reveals no matches in oe-core or meta-oe
Dec 18 23:28:22 <ant____> ah, another idea was that the packaging code could ignore the symlink during split
Dec 18 23:28:48 <ant____> ..but I could not find the symlink at all so there is no way to package those
Dec 18 23:29:32 <ant____> (I did that split by hand in the klibc recipe)
Dec 18 23:30:03 <bluelightning> ok
Dec 18 23:30:39 <bluelightning> so is this something we're missing in oe-core that we should refer to oe-classic for?
Dec 18 23:30:41 <ant____> basically if you grep for 'ln -s' or 'ln -sf' .. no interesting matches....
Dec 18 23:31:14 <ant____> yes, if we only could identify which code creates :)
Dec 18 23:31:42 <ant____> seems it is not Angstrom specific at least
Dec 18 23:32:07 <ant____> fwiw I had the symlink with generic distro as well
Dec 18 23:40:41 <bluelightning> ant____: btw I want to move the defaulting of LOGO_SIZE from zaurus.inc to linux.inc, does that sound OK?
Dec 18 23:40:50 <bluelightning> with ?= assignment though
Dec 18 23:41:11 <bluelightning> that way other machines can just work out of the box
Dec 18 23:41:13 <ant____> yes, it's ok for me
Dec 18 23:41:17 <bluelightning> cool
Dec 18 23:41:34 <bluelightning> (am working on h1940, FWIW)
Dec 18 23:42:11 <ant____> someone criticized the use of MACHINE_GUI_CLASS
Dec 18 23:46:08 <bluelightning> what would we use instead?
Dec 18 23:46:13 <bluelightning> formfactor?
Dec 18 23:47:21 <ant____> go for it for now, most machines in meta-handheld declare the old deprecated vars
Dec 18 23:52:26 <bluelightning> done
Dec 18 23:54:15 <ant____> nice
Dec 18 23:54:48 <ant____> well, I lost too much time lurking in the coide for that /boot symlink..it's now better to clone an old oe tree and verify :/
Dec 18 23:55:30 <ant____> ah, one more thing about psplash logo
Dec 18 23:55:57 <ant____> I really don't know how wrong is the yocto one as default
Dec 18 23:56:25 <ant____> I'd just prefer to see the OE one but in fact Yocto is the umbrella
Dec 18 23:57:16 <ant____> so, maybe we set a bbappend in meta-handheld where we use as well the OE kernel bootlogo?
Dec 18 23:58:00 <bluelightning> oh yeah I forgot about that
Dec 18 23:58:28 <bluelightning> we should really get it fixed in oe-core, I doubt it's still there deliberately
Dec 18 23:59:26 <ant____> maybe is a bit a sensible theme for someone. I don't want to cause flames
Dec 19 00:00:03 <ant____> ..so I outsource it to our new TSC member :p
Dec 19 00:03:16 <bluelightning> yeah I'm adding it to my todo list now :)
**** ENDING LOGGING AT Mon Dec 19 02:59:58 2011