**** BEGIN LOGGING AT Thu Oct 29 02:59:57 2009
Oct 29 05:02:51 <CIA-27> 03bzhou * r10824 10optware/trunk/make/bacula.mk: bacula: 3.0.1 -> 3.0.3
Oct 29 13:36:42 <carterb> eno, are you online
Oct 29 15:27:14 <carterb> eno?
Oct 29 15:32:22 <CIA-27> 03bzhou * r10825 10optware/trunk/ (make/sane-backends.mk sources/sane-backends/roundf.patch): sane-backends: removed roundf.patch for uclibc
Oct 29 15:36:34 <eno> carterb: see 10825
Oct 29 15:37:06 <carterb> how can i see it
Oct 29 15:37:16 <eno> (although I'm online, I'm a little busy at work)
Oct 29 15:37:25 <eno> svn up
Oct 29 15:38:01 <carterb> why did you remove it
Oct 29 15:38:27 <eno> that patch is no longer needed, and does not apply
Oct 29 15:38:33 <carterb> kk
Oct 29 15:38:53 <carterb> When will the next autobuild take place
Oct 29 15:39:24 <carterb> I sneded in a new version of snae-backends.mk yesterday, nothing happened yet
Oct 29 15:39:28 <carterb> ipkg version 5
Oct 29 15:40:10 <eno> it had build problem because of the patch
Oct 29 15:40:28 <eno> first, you only changed GIT_DATE and not GIT_HASH
Oct 29 15:40:38 <carterb> oh
Oct 29 15:40:50 <carterb> mom, I'll take a look
Oct 29 15:40:58 <eno> i made the change of GIT_HASH so it gets the new commit
Oct 29 15:41:10 <carterb> thx
Oct 29 15:41:36 <eno> keep http://cia.vc/stats/project/nslu2-linux handy
Oct 29 15:42:17 <carterb> ah, nice link
Oct 29 15:43:11 <carterb> how do you generate the hash
Oct 29 15:52:20 <eno> it's taken from sane-backends git web site
Oct 29 15:52:37 <eno> URL in the comment above SANE-BACKENDS_GIT_HASH
Oct 29 15:57:07 <carterb> is this entry in sane-backends.mk
Oct 29 15:58:19 <eno> in latest sane-backends.mk, line 33 - 39
Oct 29 16:00:48 <carterb> I cannot find a hash there, the developer tree, right?
Oct 29 16:03:43 <eno> click on any commit, you'll get the hash
Oct 29 16:05:00 <eno> git hash == commit id
Oct 29 16:05:10 <carterb> ah, that makes sense
Oct 29 16:05:17 <carterb> i see this entry :-)
Oct 29 16:05:53 <eno> BTW, we don't want to snapshot too often
Oct 29 16:05:56 <carterb> I have to set it before I change the git date?
Oct 29 16:06:11 <carterb> ok, had some big goals
Oct 29 16:06:14 <eno> change both at the same time
Oct 29 16:06:41 <eno> they should be consistent
Oct 29 16:07:24 <carterb> SANE_BACKENDS_GIT_COMMIT=92c44e905a
Oct 29 16:07:32 <carterb> I do not understand the syntax
Oct 29 16:08:20 <eno> the first >7 digits pretty much uniquely identifies the COMMIT HASH
Oct 29 16:08:52 <carterb> ok, can i choose them
Oct 29 16:09:01 <eno> read up on git and SHA1
Oct 29 16:09:17 <carterb> whats that
Oct 29 16:09:24 <eno> for 20091022, you pick "6 days ago" commit
Oct 29 16:09:44 <eno> the commit id is f2c6...
Oct 29 16:10:12 <carterb> Now the time has come again: I do not understand anything, Sorry
Oct 29 16:11:16 <eno> git is excellent, i recommend http://www.newartisans.com/2008/04/git-from-the-bottom-up.html
Oct 29 16:11:56 <carterb> ok, only if I set a new hash, the autobuilder builds the package again?
Oct 29 16:12:22 <eno> it will retrieve a new snapshot tarball
Oct 29 16:12:48 <carterb> so do I get the hash from the git snapshot?
Oct 29 16:13:12 <eno> you can, you can also browse the git website to get the hash
Oct 29 16:13:42 <eno> normally if you follow the upstream project closely, you should clone the git repo
Oct 29 16:13:48 <eno> and do "git pull" very often
Oct 29 16:13:57 <carterb> Am I right here?
Oct 29 16:13:58 <carterb> http://git.debian.org/?p=sane/sane-backends.git;a=commit;h=db1a049706785e499b546f944a8248432fc5f1ca
Oct 29 16:14:36 <eno> that's the latest
Oct 29 16:14:49 <carterb> is that the hash? fb0a60f
Oct 29 16:15:15 <eno> no, in that case the hash is db1a...
Oct 29 16:16:29 <carterb> I do not find it on that site
Oct 29 16:17:17 <carterb> http://git.debian.org/?p=sane/sane-backends.git;a=snapshot;h=fb0a60f625c07777a53edd53df2509f188744192;sf=tgz
Oct 29 16:17:30 <carterb> thats the link, so thats the hash, or not??
Oct 29 16:17:31 <carterb> fb0a60f625c07777a53edd53df2509f188744192
Oct 29 16:18:44 <carterb> sorry
Oct 29 16:18:47 <carterb> wrong site
Oct 29 16:18:48 <carterb> db1a049706785e499b546f944a8248432fc5f1ca
Oct 29 16:18:54 <carterb> thats it, right?
Oct 29 16:18:57 <carterb> no I understand
Oct 29 16:19:14 <eno> yes, you were talking about the commit "7 days ago"
Oct 29 16:19:52 <carterb> So problem with my make-file was, that autobuilder could not find the file cause of the wrong address
Oct 29 16:19:57 <eno> at this point, i'd recommend NOT follow too closely on upstream git development
Oct 29 16:20:37 <eno> you changed only GIT_DATE, so autobuilder would be retrieving old snapshot
Oct 29 16:20:44 <eno> and named it with new date
Oct 29 16:20:53 <carterb> yes, but that change was important:
Oct 29 16:20:53 <carterb> modified: backend/pixma_mp150.c
Oct 29 16:21:24 <carterb> I will not do new snapshot to often, ok!
Oct 29 16:21:59 <carterb> some guys had problems with that backend
Oct 29 16:22:52 <eno> unless you know clearly upstream fixed the problem, and you really want to get the fix
Oct 29 16:23:07 <eno> don't blindly taking snapshot, it's a bit dangerous
Oct 29 16:23:15 <carterb> yes
Oct 29 16:23:25 <carterb> You changed it for me?
Oct 29 16:23:30 <carterb> the hash I mean
Oct 29 16:23:40 <carterb> cause there is no new make-file
Oct 29 16:23:42 <eno> like the last commit, it clearly says "something (yet untested)"
Oct 29 16:23:56 <eno> i did
Oct 29 16:24:35 <eno> and with the new snapshot, roundf.patch no longer applies on uclibc target
Oct 29 16:24:43 <eno> so the build was broken for a while
Oct 29 16:25:37 <carterb> ok
Oct 29 16:25:44 <carterb> other topic
Oct 29 16:25:46 <carterb> ejabberd
Oct 29 16:25:52 <carterb> erl-ejabberd
Oct 29 16:26:03 <carterb> you updatet erlang to R13, right?
Oct 29 16:26:14 <eno> right
Oct 29 16:26:29 <carterb> cool, now it should work.
Oct 29 16:26:48 <carterb> I have the following error trying to install an ipkg-package:
Oct 29 16:26:58 <eno> AFAIK, erlang is always the latest stable release version
Oct 29 16:27:12 <carterb> Collected errors:
Oct 29 16:27:12 <carterb> Package erl-ejabberd md5sum mismatch. Either the ipkg or the package index are corrupt. Try 'ipkg update'.
Oct 29 16:27:32 <carterb> I build it on optware and tried to install from file
Oct 29 16:28:14 <eno> ipkg remembers md5sum of erl-ejabberd from the feed
Oct 29 16:28:25 <carterb> what to do
Oct 29 16:28:35 <eno> and the one you build, with the same VERSION and IPK_VERSION, has a diff md5
Oct 29 16:28:52 <eno> in your local build, you can bump IPK_VERSION
Oct 29 16:28:55 <carterb> ok, so I have to change IIPKG_Version
Oct 29 16:28:59 <carterb> kk
Oct 29 16:29:14 <eno> i usually pass it as make parameter
Oct 29 16:29:34 <eno> something like "make erl-ejabberd ERL-EJABBERD_IPK_VERSION=11"
Oct 29 16:29:50 <eno> something like "make erl-ejabberd-ipk ERL-EJABBERD_IPK_VERSION=11"
Oct 29 16:29:57 <carterb> I had an error with the last erl-ejabberd and asked the ejabberd-community for help. That was the answer
Oct 29 16:30:00 <carterb> Your new error means: "cannot open shared object file: No such file or directory", i.e. ejabberd is unable to find or load stringprep_drv.so. Are you sure it was compiled correctly?
Oct 29 16:30:15 <carterb> what package/lib is that
Oct 29 16:30:18 <carterb> stringprep_drv.so
Oct 29 16:30:36 <eno> that file sounds like should be in erl-ejabberd
Oct 29 16:31:04 <eno> optware usually only finds libs in /opt/lib
Oct 29 16:31:18 <eno> if it's not, it probably cannot find it
Oct 29 16:31:40 <carterb> ok
Oct 29 16:31:50 <carterb> I'm looking for it
Oct 29 16:32:21 <eno> right, it's under /opt/var/lib/ejabberd/priv/lib/
Oct 29 16:33:43 <carterb> hm, why do I get the error
Oct 29 16:33:44 <eno> link it to /opt/lib and see if it helps
Oct 29 16:34:03 <carterb> Im going to crosscompile it again and see if its there
Oct 29 16:34:32 <eno> if yes, set RPATH to /opt/lib:/opt/var/lib/ejabberd/priv/lib for ejabberd compilation
Oct 29 16:34:56 <eno> you probably can manipulate RPATH with ERL-EJABBERD_LDFLAGS
Oct 29 16:35:16 <carterb> yes
Oct 29 16:35:33 <eno> do a "make query-STAGING_LDFLAGS" to see how it's set to /opt/lib
Oct 29 16:36:52 <carterb> only for you to know, the change you advised me to do in postinst of sane-backends did not work,,, the _"aceof" thing. Did not write the value of $root_user nut "the string "$root_user"
Oct 29 16:36:59 <carterb> I changed it to the old version
Oct 29 16:37:16 <eno> i see that, it probably depends on /bin/sh capability
Oct 29 16:37:24 <eno> and varies from firmware to firmware
Oct 29 16:38:04 <eno> i'm surprised interpolation's not implemented in some shell
Oct 29 16:38:34 <eno> i mean /bin/sh needs to implement "$interpolation" anyway
Oct 29 16:39:09 <eno> but why stop at "", and not << (here doc)?
Oct 29 16:39:20 <eno> but thx for the test and fix
Oct 29 16:39:52 <carterb> no pro
Oct 29 18:56:13 <CIA-27> 03bzhou * r10826 10optware/trunk/make/enscript.mk: enscript: fixed build for openwrt-brcm24
Oct 29 21:23:22 <james_walmsley_U> hi
Oct 29 21:23:50 <james_walmsley_U> anyone know much about compiling the kernel from source and getting redboot to execute it?
Oct 29 21:24:23 <james_walmsley_U> I've built a cross compiler with Buildroot, and the kernel zImage to less than 1mb,
Oct 29 21:24:38 <james_walmsley_U> when I flash it, after reboot it does nothing.
Oct 29 21:24:43 <james_walmsley_U> Can anyone help?
Oct 29 21:27:03 <james_walmsley_U> hello?
Oct 29 21:27:49 <eno> james_walmsley_U: if I were you, I'd look at how existing firmware is built
Oct 29 21:28:29 <eno> expecting ppl to explain from start to finish is too much
Oct 29 21:29:11 <james_walmsley_U> no, I'm not expecting a start to finish thing, I already ported Linux to a few different boards, maybe I just have the wrong toolchain compiler
Oct 29 21:29:27 <eno> start from http://www.nslu2-linux.org/wiki/Development/MasterMakefile
Oct 29 21:29:53 <eno> build one existing firmware, and see what is the diff
Oct 29 21:30:14 <james_walmsley_U> yeah, I did that already, but it just does everything for u, and runs a few bitbake files, but it doesn't explain what it does.
Oct 29 21:30:36 <james_walmsley_U> I wanted to get an executable kernel onto the flash, that will kernel panic at no filesystem
Oct 29 21:31:30 <eno> you'll need to read the relevant bb file then
Oct 29 21:31:54 <eno> or, in the case of openwrt, some makefil
Oct 29 21:31:56 <eno> e
Oct 29 21:33:07 <james_walmsley_U> hmm, 
Oct 29 21:33:39 <james_walmsley_U> where exactly is the bb file for openslug, which contains the cross compiler build, and kernel build
Oct 29 21:34:31 <james_walmsley_U> the mastermakefile is nice if you want everything done for you, but its not exactly nice for a developer to have to go through trying to find out how openslug build everythign
Oct 29 21:35:46 <eno> i don't know much OE myself, but I guess you'll need to do a lot of searching
Oct 29 21:36:35 <eno> they definitely have separate .bbclass file for toolchain, kernel, etc.
Oct 29 21:38:47 <eno> maybe openembedded/classes/nslu2-image.bbclass
Oct 29 21:40:49 <james_walmsley_U> thanks, I'll take a look
Oct 29 21:41:10 <james_walmsley_U> I used a compiler compiled for oabi, do I need to be eabi?
Oct 29 21:41:24 <eno> and for toolchain, probably openembedded/recipes/gcc
Oct 29 21:42:19 <eno> i don't think you have to use eabi
Oct 29 21:42:50 <eno> there're nslu2 firmware predates eabi
Oct 29 21:47:03 <eno> see also openembedded/conf/distro/slugos.conf and openembedded/conf/distro/include/slugos.inc
Oct 29 21:53:47 <james_walmsley_U> just to confirm the chip is an xscale?
Oct 29 21:54:33 <eno> Processor       : XScale-IXP425/IXC1100 rev 1 (v5b)
Oct 29 22:00:48 <james_walmsley_U> my compiler is called armeb-linux-uclibc-gcc
Oct 29 22:01:15 <james_walmsley_U> but the openembedded one is called arm5vteb-linux-gnueabi
Oct 29 22:20:52 <Reedy> eno, hey, did you happen to see what i said about the master make file yesterday?
Oct 29 22:21:39 <eno> Reedy: i did, rwhitby is the owner of MasterMakefile
Oct 29 22:22:43 <Reedy> ahh
Oct 29 22:24:28 <eno> Reedy: send email to nslu2-developers
Oct 29 22:27:45 <CIA-27> 03bzhou * r10827 10optware/trunk/make/zile.mk: zile: 2.3.12 -> 2.3.13
Oct 29 22:27:47 <CIA-27> 03bzhou * r10828 10optware/trunk/make/git.mk: git: 1.6.5.1 -> 1.6.5.2
Oct 29 22:28:33 <Reedy> eno, done
Oct 29 22:29:30 <Reedy> Need to have a poke at setting my optware env up again tommorow night... It was messing about last night
Oct 29 22:30:02 <eno> Reedy, let me know if you need any help
Oct 29 22:30:06 <Reedy> thanks :)
Oct 29 22:30:17 <Reedy> Wouldnt let me make ipkg-opt, giving a permissions error, but nothing more, trying to make which led to a directory owned by an unknown group/user
Oct 29 22:30:44 <Reedy> Fixed that, and it spent a while going away (it actually downloaded the toolchain then, hadn't tried for ipkg-opt)
Oct 29 22:30:47 <Reedy> And then errored with something else
Oct 29 22:31:00 <eno> for which target?
Oct 29 22:31:09 <Reedy> just nslu2 (so unslung?)
Oct 29 22:31:37 <eno> (I haven't build nslu2(unslung) toolchain for a while)
Oct 29 22:31:41 <Reedy> hmm
Oct 29 22:32:03 <Reedy> Hmm, found it
Oct 29 22:32:05 <Reedy> eno, <Reedy> drwxr-xr-x 7 15622  5000 12288 2009-10-29 00:59 crosstool
Oct 29 22:32:18 <Reedy> was the problem that was causing it to bork for which originally
Oct 29 22:32:24 <Reedy> I'd just followed the wiki guide before
Oct 29 22:33:19 <Reedy> (ie make directories etc)
Oct 29 22:33:31 <eno> what distro and version for your build machine?
Oct 29 22:33:41 <Reedy> it was ubuntu server 9.04
Oct 29 22:34:10 <eno> 32bit or 64bit? (not sure if that matters)
Oct 29 22:34:15 <Reedy> I tried both
Oct 29 22:34:30 <eno> but I do have access to some 9.04 box
Oct 29 22:34:47 <Reedy> 64bit was causing errors when i was looking at the master makefile (prompting the messages yesterday), so i whacked a 32bit vm on to see
Oct 29 22:36:40 <eno> oh, the psyco one
Oct 29 22:36:53 <eno> it's only needed for OE bitbake
Oct 29 22:37:44 <Reedy> hmm
Oct 29 22:38:24 <eno> just launched a build of nslu2 toolchain
Oct 29 22:38:29 <Reedy> gcc-3.3 is rather old... (3.4 seems available)
Oct 29 22:38:37 <Reedy> More weird ubuntu and debian were quite different
Oct 29 22:39:10 <eno> the point of unslung firmware is to be bug/feature compatible with linksys
Oct 29 22:39:29 <eno> toolchain version really matters here
Oct 29 22:39:35 <eno> and cannot be changed easily
Oct 29 22:39:37 <Reedy> hmmm
Oct 29 22:40:12 <eno> there're cases that binary contains reference to GCC_* symbols
Oct 29 22:40:53 <eno> gcc-3.3 is old
Oct 29 22:46:35 <Reedy> eno, i'll have another look tommorow when i've got some more time (got a lovely exam in the morning)
Oct 29 22:47:52 <eno> alright, my build is still going on, but so far it seems fine
Oct 29 22:48:23 <Reedy> hmm, no initial permissions error?
Oct 29 22:48:31 <eno> no
Oct 29 22:48:41 <Reedy> weird
Oct 29 22:51:41 <Reedy> what are you building?
Oct 29 22:54:23 <eno> just the toolchain
Oct 29 22:54:53 <eno> it's done
Oct 29 22:55:22 <eno> and "make which-check" is fine
Oct 29 22:55:48 <Reedy> I'll have a look tommorow :)
Oct 29 22:55:49 <Reedy> thanks
Oct 29 22:55:51 <eno> oh, I just realized the machine I'm using is 8.04
Oct 29 22:55:55 <Reedy> heh
Oct 29 22:55:59 <Reedy> 9.10 is out ;
Oct 29 22:55:59 <eno> sorry
Oct 29 22:56:00 <Reedy> ;)
Oct 29 22:56:43 <eno> my 9.04 machine at home is shutdown
Oct 29 22:56:56 <eno> will try that when I get home
Oct 29 22:56:56 <Reedy> alright, no rush :)
Oct 29 22:57:01 <Reedy> Thanks
Oct 29 22:58:30 <eno> FYI, the official build machine is also on 8.04
Oct 29 22:59:09 <Reedy> which is LTS... So that figures
Oct 30 00:31:28 <CIA-27> 03bzhou * r10829 10optware/trunk/make/zile.mk: zile: reverted to 2.3.12
**** ENDING LOGGING AT Fri Oct 30 02:59:58 2009