**** BEGIN LOGGING AT Fri Feb 05 02:59:56 2010
Feb 05 06:09:55 <CIA-10> 03bzhou * r11167 10optware/trunk/ (make/openssh.mk platforms/packages-syno-i686.mk): openssh: without-stackprotect on syno-i686
Feb 05 06:34:11 <CIA-10> 03bzhou * r11168 10optware/trunk/make/golang.mk: golang: 20100127 -> 20100204
Feb 05 07:43:42 <CIA-10> 03bzhou * r11169 10optware/trunk/platforms/packages-syno-i686.mk: openssh: without-shadow on syno-i686
Feb 05 12:04:50 <rwhitby> eno: I'm building the dependencies of optware-devel on the ds1010+ as a test of optware-devel on that platform :-)
Feb 05 12:17:34 <CIA-10> 03rwhitby * r11170 10optware/trunk/make/optware-devel.mk: Added tcl to optware-devel
Feb 05 12:21:53 <CIA-10> 03rwhitby * r11171 10optware/trunk/make/tcl.mk: Symlink tclsh8.4 as tclsh
Feb 05 12:26:17 <Garoo> hello
Feb 05 13:05:37 <CIA-10> 03rwhitby * r11172 10optware/trunk/make/cyrus-sasl.mk: Additional targets to help self-hosting testing
Feb 05 13:10:48 <CIA-10> 03rwhitby * r11173 10optware/trunk/make/e2fsprogs.mk: Additional targets to help self-hosting testing
Feb 05 13:42:47 <CIA-10> 03rwhitby * r11174 10optware/trunk/make/openldap.mk: Additional targets to help self-hosting testing
Feb 05 13:49:18 <jomaster> rwhitby: shouldn't you add the lib-targets, you add, to the PHONY-target and also to the clean-targets?
Feb 05 13:49:44 <rwhitby> jomaster: indeed I should, yes.
Feb 05 13:52:19 <rwhitby> but for the purpose for which they are being added (ipkg list_installed | awk | xargs make ...), it's not essential.
Feb 05 13:56:00 <jomaster> ah ok, i thought that you like to use this targets as depencies.
Feb 05 13:57:33 <rwhitby> no, I'm testing self-hosting of optware on a synology ds1010+ (dual core atom)
Feb 05 15:15:54 <eno> rwhitby: cool
Feb 05 15:16:17 <rwhitby> eno: perl fails
Feb 05 15:16:28 <rwhitby> pp.c:(.text+0xa5a4): undefined reference to `floor'
Feb 05 15:16:59 <rwhitby> gcc fails:
Feb 05 15:17:02 <rwhitby> The directory that should contain system headers does not exist:
Feb 05 15:17:03 <rwhitby>   /usr/include
Feb 05 15:17:04 <rwhitby> make[4]: *** [stmp-fixinc] Error 1
Feb 05 15:17:30 <rwhitby> pkgconfig fails:
Feb 05 15:17:31 <rwhitby> Can't exec "aclocal-1.9": No such file or directory at /opt/share/autoconf/Autom4te/FileUtils.pm line 326.
Feb 05 15:17:54 <rwhitby> python fails:
Feb 05 15:17:54 <rwhitby> make: *** /opt/src/optware/syno-i686/builds/python25/buildpython25: No such file or directory.  Stop.
Feb 05 15:18:02 <rwhitby> everything else builds fine
Feb 05 15:18:04 <eno> on ds1010+, do you use optware as native compilation or cross?
Feb 05 15:18:19 <rwhitby> I installed optware-devel
Feb 05 15:18:28 <rwhitby> and did:
Feb 05 15:18:45 <rwhitby> ipkg list_installed | awk '{ print $1 }' | xargs make 
Feb 05 15:20:32 <eno> k, will look at it during the weekend
Feb 05 15:20:59 <eno> /usr/include is understandable, because we put headers in /opt/include
Feb 05 15:21:11 <eno> Friday morning here
Feb 05 15:24:52 <rwhitby> being able to complete optware-devel self-hosted on the device will be a good milestone
Feb 05 15:25:02 <rwhitby> it builds quite quickly too
Feb 05 15:25:33 <eno> yeah I agree, will work towards that
Feb 05 15:42:01 <CIA-10> 03bzhou * r11175 10optware/trunk/platforms/packages-syno-i686.mk: openssh: reverted --without-shadow change on syno-i686
Feb 05 16:05:16 <CIA-10> 03bzhou * r11176 10optware/trunk/ (make/erlang.mk platforms/packages-syno-i686.mk): erlang: enable-smp-support on syno-i686
Feb 05 16:09:33 <jomaster> i don't understand the reason for selfhosting.
Feb 05 16:12:10 <jomaster> because some packages require specific versions of some tools (like aclocal-1.9), i proposed last year to build required tools as host-stage like openwrt do.
Feb 05 16:29:31 <eno> jomaster: selfhosting itself is just a milestone, reducing host tools may very well be the way to achieve that
Feb 05 16:32:46 <jomaster> perhabs there should be a list of all required host tools (and their required versions)
Feb 05 16:35:50 <jomaster> for selfhosting it is imortant, because optware-devel should have it.
Feb 05 16:36:42 <jomaster> ... have all this tool :-D
Feb 05 16:41:03 <jomaster> eno: selhosting only works if the toolchain is available as sources, isn't it? all targets with binary toolchains can't do so.
Feb 05 16:41:23 <jomaster> on how many targets optware-devel is available?
Feb 05 17:14:27 <eno> jomaster: well, there might be different meanings of selfhosting. In this context, it simply means that without other distribution packages, just using optware-devel, we will able to create a optware devel environment
Feb 05 17:15:30 <eno> so, no specific toolchain source is needed, just standard gcc/binutils source
Feb 05 18:02:34 <jomaster> eno: why optware-devel depence on crosstool?
Feb 05 18:08:16 <eno> jomaster: optware-devel depends on native toolchain if one is available
Feb 05 18:08:29 <eno> on some platforms, it's crosstool-native
Feb 05 18:08:54 <eno> on some other, it's gcc + binutils
Feb 05 18:09:05 <jomaster> what is the different between compiled gcc and binutils and nativ toolchain?
Feb 05 18:09:34 <eno> gcc + binutils are basically native toolchain
Feb 05 18:10:26 <jomaster> so nativ toolchain is only a detour over crosstool?
Feb 05 18:10:59 <eno> depends on how native toolchain was built, packages are named differently
Feb 05 18:11:34 <eno> on some platforms, the cross toolchain was built using crosstool
Feb 05 18:11:55 <eno> native toolchain was also based on crosstool build environment
Feb 05 18:12:10 <eno> that's why it's called crosstool-native
Feb 05 18:13:18 <jomaster> understood
Feb 05 18:14:43 <eno> if we use optware/make/binutils.mk and gcc.mk to build a native toolchain, the result packages are just called binutils and gcc
Feb 05 18:15:15 <jomaster> where is the problem?
Feb 05 18:15:18 <eno> optware-devel tries to hide this detail
Feb 05 18:16:04 <eno> what do you mean "where is the problem?" ?
Feb 05 18:16:17 <jomaster> 'if we use optware/make/binutils.mk and gcc.mk to build a native toolchain, the result packages are just called binutils and gcc'
Feb 05 18:16:41 <eno> no problem
Feb 05 18:16:44 <jomaster> that is normal on host, isn't it?
Feb 05 18:17:18 <eno> it varies from target to target
Feb 05 18:18:35 <jomaster> so, if gcc, libc-dev and binutils works, we have a nativ toolchain. why is optware-devel disabled on fsg3v4?
Feb 05 18:19:53 <eno> is it? I did not know that, feel free to correct
Feb 05 18:22:03 <eno> it was disabled because crosstool-native was broken
Feb 05 18:22:39 <eno> I don't see libc-dev/gcc/binutils in SPECIFIC_PACKAGES of platforms/packages-fsg3v4.mk
Feb 05 18:22:40 <jomaster> didn't try or use gcc or optware-devel on fsg3v4 but know there is following sentence on packages-fsg3v4.mk: 'crosstool-native is not available (and therefore neither is optware-devel)'
Feb 05 18:23:37 <jomaster> they aren't standard. i only look for them at brocken packages.
Feb 05 18:23:40 <eno> it's not a simple as "use gcc", you need to build and test "libc-dev/gcc/binutils/gcc"
Feb 05 18:24:16 <eno> that would require device access, so you're probably the best person to make it happen
Feb 05 18:33:12 <jomaster> so, only try to build gcc, binutils and libc-dev and then test them on fsg3v4?
Feb 05 18:41:25 <eno> jomaster: yes
Feb 05 18:42:23 <jomaster> ok, than i will try to remove all reported bugs along next weeks. perhabs i will find a night for this :-D
Feb 06 01:33:40 <CIA-10> 03bzhou * r11177 10optware/trunk/make/nginx.mk: nginx: 0.7.64 -> 0.7.65
**** ENDING LOGGING AT Sat Feb 06 02:59:56 2010